The World Economic Forum’s Risks Report 2023 ranks failure to mitigate and adapt to the climate crisis within its top two priorities globally, for both the near and long term. As climatic change leads to a rise in erratic weather patterns, the increase in storm intensity and unpredictable surges have become a significant challenge for global hydropower infrastructure. Surges cause severe damage to turbines, spillways, and ancillary infrastructure, and increased difficulty in accurately predicting these events leaves operations teams unprepared to respond effectively, leading to severe infrastructure damage and compromised energy production.
In 2017, California’s Department of Water Resources issued a mandatory evacuation order for 188,000 residents living below the Oroville Dam, fearing catastrophic failure. A series of unseasonable storms had caused Oroville Lake to rise rapidly, exposing maintenance vulnerabilities in the dam’s spillway and severe structural problems.
A recent response to the climate crisis comes from Statkraft, Europe’s largest renewable energy producer, which has announced a capital investment programme amounting to €700 million to shore up its hydropower assets against the impact of increasing storm surges. Large scale capital investment programmes like these demonstrate the industry’s ongoing commitment to building resilience in ageing infrastructure against the climate threat.
But, at present, tens of thousands of dams worldwide are ageing, and ill-equipped to withstand the increased frequency and intensity of near-full or spillover events and overtopping caused by climate volatility. According to the Association of Dam Safety Officials, overtopping accounts for 34% of all dam failures. As reservoirs approach capacity, their structural integrity is threatened, straining spillways and surrounding structures.
Meteorological forecasting
Hydropower generation depends on accurate water forecasting for capacities and flows to inform both the design and construction of dams, as well as long term operations. But rapidly changing meteorological conditions mean that traditional models of water forecasting can no longer accurately account for an increase in volatility in heavy rainfall patterns. Subsequent inaccurate predictions, therefore, may lead to poor construction design, which has been shown to increase costs, and present potential safety hazards over time, as well putting operational decision making under pressure and leaving current dam safety procedures wanting.
Set against the emerging challenges in forecasting, nevertheless, are the responsibilities that dam owners have to human life, property and the environment in their operations. Regulatory requirements for hydropower operators demand the mitigation of spillover events to ensure safety and compliance. The safety breach at the Oroville Dam triggered an extensive review of dam safety practices across the US, with the Federal Energy Regulatory Commission (FERC) driving significant changes to safety standards and regulations. Balancing energy production and regulatory compliance has always presented a significant burden on the operator, and emerging gaps in water forecasting present further challenges to operators looking to refine the decision making between when to produce power, while balancing compliance with social, ecological and environmental regulation.
To protect third-party property, operators must conduct regular inspections, maintenance, and safety evaluations to mitigate the risk of flooding that can damage private properties downstream. Compliance requires maintaining specific stream flows to preserve local ecosystems. Erratic water releases can disrupt habitats, endanger species, and degrade water quality. Recreational activities, which rely on stable predictable water levels, such as fishing, may also be impacted. Regulatory frameworks mean that operational needs must be balanced with consideration for activities in the area and accurate water forecasting plays a key role in achieving this.
The bridge to a smarter future
AI learning tools have already demonstrated their capability in comparable meteorological contexts where they have outperformed the accuracy of established forecasting models more than 90% of the time, using smaller, more widely available hardware.
AI models provide high-resolution forecasting, which works by collecting weather from different sources to simulate atmospheric conditions at a fine scale. These targeted forecasts offer the opportunity for more efficient resource allocation and water management. Precedents for translating this performance exist and show clear benefits as well as the rapid return on AI investment available for the hydropower industry.
AI forecasts are starting to see uptake amongst the major traditional forecasting organisations – such as the European Centre for Medium-Range Weather Forecasts (ECMWF) – as the pre-eminent technology firms unveil their own proprietary technology.
Managing flood and storm events
While the adoption of AI for weather forecasting is still embryonic, hydropower operators are embracing the technology for the management of flood and storm events. This advanced water forecasting provides insights which enable operators to implement proactive measures that reduce infrastructure damage and downstream flooding. AI can generate forecasts at the basin watershed or plant level, providing targeted insights for precise water management that understand local hydrological dynamics and inform accurate predictions. To date, AI has already been used to manage water resources in complex watersheds with diverse climatic conditions, resulting in significant reduction in operational risks related to water variability, achieving improvements in forecast precision and decision-making.
Protecting construction and O&M
Alongside the management of storm surge events, operators are continually working to safeguard their ongoing construction projects, as well as routine operations and maintenance (O&M), from unexpected water level changes. Confidence in the reliability and accuracy of AI-based predictions enables the scheduling of works to occur during safer periods, which reduces risk and enhances safety. AI water forecasting applied in the Pacific northwest region has demonstrated the ability to optimise reservoir management and improve energy production and carries the additional benefit of enhanced environmental compliance. In Europe, where reservoirs have smaller capacities, the windows to act on insights related to changeable water volumes are compressed, and AI’s real-time processing improves decision making capabilities.
AI has also been used successfully in the identification of suitable locations for dam construction by analysing environmental, geographical, and social data. The Nature Conservancy, along with computer scientists, has used AI to evaluate hundreds of existing and proposed hydropower sites across the entire Amazon basin, to find sites with minimal ecological disruption and eliminating the expense of moving or cancelling operations.
Operators must comply with regulations like those set by FERC, which require maintaining specific reservoir levels. AI-based tools help by simulating different meteorological models and predicting water level changes using Machine Learning algorithms. These methods allow operators to manage water flows and quality effectively, ensuring compliance.
A sustainable future
In its 2024 World Hydropower Outlook, the International Hydropower Association advocates for harnessing data analytics, machine learning, and predictive maintenance algorithms. Adopting these tools gives “operators deeper insights into plant performance, water management, and condition monitoring.”
AI-based water forecasting and modelling are leveraging advanced machine learning algorithms, real-time data processing, and climate adaptation models, to provide precise and actionable insights that support efficient, safe, and sustainable hydropower management. These technologies support water level predictions, scheduling optimisation, and safeguard regulatory compliance for the hydropower industry, ensuring a route to a sustainable future for hydropower.
