Optimisation methods for water management
Expert(s) |
Bernhard Becker
How do we best make use of our existing water infrastructure? And how do we help water managers balance the various interests they face in their daily work? Optimisation techniques for operational water management have now been assembled by Deltares for the first time in the overview paper 'Optimisation methods in water system operation'. The authors have compiled optimization techniques that they have been using in practice and the paper is intended for (prospective) water managers. At the same time, the authors invite fellow scientists to further develop optimisation techniques specifically for water management.
Water management must consider an increasing number of social challenges: water safety, energy transition, nature restoration, drinking water quality and economic progress. For managers of polder pumping stations and sluices, this means they must weigh up many interests in their daily work. Pumping water away in lowland areas, for instance, comes with CO2 emissions, but is necessary for dry feet. In other countries, energy is generated with flowing water; how can energy yield from a hydropower plant be increased while meeting environmental requirements such as fish passability?
Water managers often have to balance between different and sometimes conflicting interests, such as flood protection and storing water for summer drinking water supply. Optimisation models help prioritise and resolve conflicting purposes and make better use of our existing water infrastructure.
Bernhard Becker, reservoir management expert and product owner RTC-Tools software
Future challenges
Hydraulic models are common practice for water managers in designing studies and for supporting operational decisions. These models mainly show the effects of a certain control on the water system. Optimisation models are more advanced as they show exactly how to control to achieve certain goals, such as: keeping dry feet through optimal use of pumps and sluices, low energy costs (or low CO2 emissions) for pumping stations, maximum energy generation from hydropower or 'fish-friendly' steering of pumping stations or turbines.
Optimisation techniques help to make the use of existing water infrastructure future-proof by, for example, reducing energy consumption for water management with a positive impact on energy transition, increasing water safety levels through better use of storage space in reservoirs and through smart control of polder pumping stations.
Better use
By listing optimisation techniques and indicating which technique is best to use when, based on practical experiences, the authors of the paper 'Optimisation methods in water system operation' aim to disseminate knowledge to current and future water managers. Meanwhile, specially developed software has become available for modellers to build their own optimisation models, this will also ensure that the knowledge from the paper can be used in practice. Deltares also offers software for developing optimisation models: RTC-Tools. RTC-Tool models are already in daily use at several water managers.
The authors' aim is to help make better use of existing water infrastructure. New locks, pumping stations or pumps are expensive, and, in addition, climate change is causing increasingly erratic conditions and more frequent flooding. Under these conditions, water managers must be able to continue to ensure water safety.
Model-based control using optimisation models is daily practice in industry. However, operational water management presents specific challenges: a large system with multiple hydraulic structures, changing hydrological conditions with extremes such as flooding or drought, and both linear and non-linear equations with continuous and discrete elements, to name a few.
Klaudia Horváth, optimisation modeller model predictive control at Deltares
Uncertainties
Flooding requires anticipation of future events, and many uncertainties are involved in water management. Specific optimization techniques can take these uncertainties into account.
We have techniques ready to deal with hydrological uncertainty, but there are more sources of uncertainty in operational water management. For example, energy prices and energy and water demand also bring uncertainty.
Jesús Andrés Rodríguez-Sarasty, operations research expert stochastic optimisation at Deltares
Linear
To control structures optimally, we need to integrate the same equations which are in our hydraulic models into the optimisation models. These equations are often non-linear. However, optimisation techniques work best with linear equations.
A core element of an optimisation model is the so-called optimisation solver. For different types of optimisation problems (linear, non-linear, continuous, discrete), there are also different solvers. With the paper, the authors aim to make developers of optimisation solvers understand where the challenges lie in designing optimisation models for water management and invite them to find solutions to them.
If we can show the necessity to solve non-linear equations in optimisation models, I am confident that the operations research community will accept the challenge and further develop solvers for non-linear optimisation problems.
Ailbhe Mitchell, operations research expert mathematical optimisation at Deltares
The paper contains practical experiences of the latest developments around optimisation techniques such as non-linear optimisation, optimisation under uncertainty and the use of optimisation models in decision support systems.
Mathematical optimisation should be standard on the academic curriculum of modellers and among users of optimisation models.
Co-author Caroline Jagtenberg, Ass. Prof. operations research at VU Amsterdam
