Project: Bringing reservoir operation into practice
Supervisors: Dr Francesca Pianosi and Professor Thorsten Wagener
Worldwide, the most common purpose of reservoirs is to deliver a reliable water supply, commonly for domestic and industrial use, irrigation or thermal plant cooling. Reservoirs enable a reliable extraction of water from a river at a rate between the river’s minimum and mean flow where, without a reservoir, only the river’s minimum flow could be reliably yielded. Other common functions beyond water yield also exist, including hydropower protection and flood control. These are the primary purposes to justify the construction of large (i.e. above 15 m) dams. Globally, 90% of large dams are designed towards these, or a combination of these, aims. In England the purpose is predominantly for supply with over 75% of reservoir capacity owned by water companies. A further 15% is owned by the Environment Agency for flood protection and conservation.
Optimization methods are essential tools to inform efficient reservoir design. The primary task of the design process is to determine the smallest capacity for which the reservoir can reliably yield its demand. Efficient design includes understanding and modelling how an asset will be operated otherwise one cannot simulate the reservoir’s performance.
If extensive water infrastructure already exists, reservoir operation optimization delivers other benefits. Due to the intrinsic difficulty of forecasting great distances into the future, dams often become unsuited to the conditions first envisaged in their design. Rather than expending capital on constructing new dams, optimization methods can tell us how the reservoir’s operation might be altered to fit within new conditions or how to efficiently operate a network of supplies for increased reliability.
A range of mathematical optimization algorithms and techniques has been applied to the reservoir operation problem.
The first aim of this research is to propose a classification of reservoir optimization approaches by focusing on the formulation of the water management problem rather than the optimization algorithm type. We believe that decision makers and operators will find it easier to navigate a classification system based on the problem characteristics, something they can clearly define, rather than the optimization algorithm. Part of this study includes an investigation regarding the extent of algorithm uptake and the possible reasons that limit real world application. This will help researchers to improve the accessibility of their work to practitioners and increase uptake.
Our second aim is to develop new algorithms to take advantage of changing computing and sensing technology. In the past, little real-time data has been available to reservoir operators. Resultantly few operation optimization algorithms utilize it, preferring instead to use static policies based on historical inflow records. Few methods that do utilize real-time data typically assume simplistic flow regimes and are not applicable to systems of more than three reservoirs. However, water suppliers are under a constant drive to reduce environmental impact and increase yields in a changing climate, and therefore they are moving towards better networked water systems. We propose using heuristic optimization and machine learning to create scalable real-time operation algorithms that will take full advantage of increasingly available real-time data.
Barney is in the water and environmental engineering group at University of Bristol. Before starting the WISE CDT he graduated from University of Bristol civil engineering with a first and the ICE student prize.
I’m looking to interview or receive information from those involved in the reservoir operation process. My first aim is to determine the degree of control operators have over the reservoir’s releases and other constraints placed upon operation. My second aim is to assess the extent of formal and informal decision making involved in the process of long and short term operation. My third aim is to assess the degree of understanding, perception and uptake of operation optimization methods.
If you feel you can offer help or advice towards any of these collaborative aims please don’t hesitate to contact me at email@example.com.
See Barney’s poster presented at the European Geosciences Union (EGU) General Assembly 2016 in Vienna, Austria.