- Published 2011 in ICNSC

For the operation of hydropower plants, the following often contrary – requirements have to be considered: safety of navigation, intensive use of hydropower, flood protection, water demand for industrial and irrigation purposes and minimization of the number of actuator operations (turbines, weirs). Today the automated control of a reservoir or a cascade of reservoirs is of increasing importance. On the one hand, the control system has to meet the above-mentioned requirements with increasing demands in control performance and high reliability of operation. On the other hand, there is the complex, highly nonlinear, unsteady and locally distributed hydraulic behaviour in the reach as well as the variable operation conditions of the barrage with changing requirements for discharges via turbines and weirs. The main influence on the hydraulic behaviour is determined by the operation of locks, the demand-driven production of electric energy, flow into or out of the reach, rain, snowmelt, etc. For the sake of reliability, each reservoir is controlled by a local controller. In the case of a cascade of reservoirs and especially if a considerable reservoir volume is available, a central water management system in addition to the local controllers is favourable. An upstream water gauge with periodical discharge measurements is needed to provide a forecast of discharge in the controlled river reaches. A specifically adapted optimization tool helps to meet requirements like power modulation for the optimization of energy production or prevention of the amplification of discharge variations. The method used to design the local controller and the central water management system is a hydrodynamic-numerical model coupled with control algorithms and optimization tools, all interacting from one calculation time step to the next. The paper describes a numerical method to design the structure and to optimize the parameters of the central water management system, taking into account its requirements with minimum design and realization efforts. It has been applied successfully in several projects in research and practice. Case studies of the rivers Rhine and Moselle show its practical application. SIMULATION OF THE HYDRAULIC PROCESS IN A RESERVOIR Prior to realizing an automated control of a barrage, investigations have to be carried out in order to evaluate the influence and effects of the automated control on the reservoir. These investigations refer to the interaction between the operation of the hydropower plants and weirs and the hydraulic situation within the reservoir. Figure 1: Aerial photo of a barrage and longitudinal section of an impounded river system (Moselle), [4] To describe the hydraulic process in a reservoir, a 1-dimensional unsteady hydrodynamic-numerical (HN-) model is used. The basic equations of the model are the classical equations of Saint Venant for 1-dimensional open channel flow (continuity and dynamics). The geometry of the reach is described by cross-sections. At the locations of hydropower plants and barrages, unsteady boundary conditions of discharge, water level and rating curves are defined. Special elements like weirs, gates, storage basins, bridges, siphons and looped or meshed river systems can be integrated [1]. The independent variables in this equation system are time t and place x, whereas the dependent variables are discharge Q(x,t) and wetted area A(x,t) or water depth y(x,t). This equation system is solved using the Preismann implicit method. One of the main tasks of an engineer is to model a real process as exact as necessary and not as exact as possible. As detailed information, such as velocity distribution, is not necessary, there is no need for 2or even 3-dimensional HNmodelling. The costs of control design can therefore be kept low by using a 1-dimensional HN-model [1,2]. Nevertheless a purely hydrological (reservoir) model is not detailed enough as the modelling of the flow in the river is indispensable. The relevant variables for the controller are water level and discharge at certain stations. The hydraulic model is embedded in a MATLAB/Simulink-Environment which facilitates the detailed design of the control algorithms and offers optimization tools [1]. All elements of the simulation – hydraulic process and control structures – interact from one calculation time step to the next. AUTOMATED CONTROL WITH LOCAL CONTROLLERS In order to fulfil the above-mentioned, often contrary requirements, the choice of an intelligent operation strategy for a barrage depends on the analysis of the hydraulic situation in the reservoir. An a priori classification of the reservoir with respect to the ratio of wave propagation time TL and retention time TR is used to choose an adequate strategy [2]. The wave propagation time TL is defined as the period of time between a variation of discharge (e.g. the inflow into a reservoir) and the resulting variation of water level at a reference point (e.g. the headwater level of a barrage). The retention time TR is the time interval needed to take in/out the retention volume at a given change in discharge with the retention volume being the volume difference corresponding to two different steady state discharges at the same headwater level. The wave propagation time TL as well as the retention time TR depends on the discharge and the headwater level. The closed loop level control is defined as a barrage operation strategy where the water level at a reference point in most cases the headwater level at the barrage is fixed to a constant value. Information about the inflow into the reservoir is not taken into account. For a reservoir with TR > TL, the standard closed loop level control often is the strategy of choice, because the water level can be kept at the reference value without any amplification of oscillations of the inflow hydrograph [2]. For a reservoir with TR < TL, maintaining a constant water level using a standard closed loop level control leads to an amplification of inflow oscillations and therefore is an improper operating method. For these reservoirs, the only way to avoid amplification of inflow oscillations is to allow a slight variation of the headwater level and to give information about incoming discharge to the control unit for the outflow ahead of time (before TL). In most cases, this is done using the feedforward control to controller output. The inflow information is given time shifted, but directly as a set value to the outflow (feedforward control). An additional cascaded proportional-integral (PI-type) level controller is used to keep the water level within a given tolerance even if the feedforward control operates with incorrectly measured inflow values. In Germany this type of control of barrages is often called OW/Q-method [2]. For both, standard closed loop level control and feedforward control, the parameters of the PI-level controller and the other parameters need to be adapted to the specific reservoir behaviour. As a starting point for the parameter optimization of the automated control, the following hydraulic values of the reservoir have to be known: wave propagation velocity times and volumes of retention impounded surface These values are not constant but depend on the discharge and the headwater level. The state of the reservoir varies widely between high flow and low flow, which has to be considered in the optimization process of the control system. At many impounded river systems in Germany, the allowance for headwater level variation is rather small. Hence the main goal of the local controllers is to maintain the water level near a constant reference value. Provided an adequately chosen structure and parameter set, local controllers can perform this task well without any amplification of discharge oscillations. Unfortunately, in practice there are many local controllers which do not avoid amplification. If, in an impounded river system, there is an allowance for headwater level variations which makes a considerable volume available for water management, superordinate strategies of water management are favourable to make the most of the resulting possibilities. This is described in the following section. STRUCTURE OF THE CENTRAL WATER MANAGEMENT SYSTEM The intent followed by a central water management system can be, for example, the amelioration of navigation conditions or the optimization of energy production. For energy production, it is beneficial to work with power modulation, rising the energy production during peak times of energy demand. Therefore it is necessary to use the full capacity of the reservoir within the concession limits, considering at the same time an ideal filling and emptying of the available volume. A specially adapted optimization tool can achieve this [3,4]. Another important field for the use of central water management systems is the minimization of discharge oscillations. In many rivers with cascades of barrages like the Moselle (see Figure 1), there are relevant unnecessary oscillations of discharge due to inappropriate controllers at upstream barrages. These discharge oscillations can cause severe problems e.g. for navigation conditions, especially in times of low water. Therefore, a central water management system should be superordinated. In the following, the different elements of a central water management system are described using the example of the upper Moselle. The considered part of the Moselle is located near the Franco-German border and consists of the reservoirs Palzem, Grevenmacher and Trier. Together they form a stretch of the Moselle of aprox. 46 km in length. The department of Hydraulic Engineering and Water Resources Management, University Kassel, on behalf of the Federal Waterways Engineering and Research Institute, carried out a research project on this stretch. The goal of the project was the development of a central water management system which would allow to utilize the available water management volumes of the three reservoirs to attenuate the discharge oscillations [4]. From measurements of the upstream water gauge Hautconcourt, the inflow to Palzem is predicted. This forecast is used to define an ideal discharge hydrograph for the downstream end of the considered stretch. The possibility to reach this ideal discharge is verified and the final discharge for each barrage is determined by the central coordinator in cooperation with the local controllers. The structure of this two-layer-concept with local controllers and a central coordinator is represented in Figure 2. The central coordinator communicates with the local controllers, getting information about discharges and water levels. Additionally, it is fed with information about the current inflow into the most upstream reservoir considered and a forecast for future inflow. Considering all this information, the central water management system calculates the ideal water management strategy for all barrages and imposes the necessary changes of discharge (∆Q) and water level reference value (∆H) to the local controllers. This two-layer concept provides a redundancy and guarantees the function of the system even if the central water management system fails. central water management local controller local controller local controller ∆Q, ∆H ∆Q, ∆H ∆Q, ∆H inflo w (Q ) inflow forecast (Q) Figure 2: Structure of a central water management system communicating with the local controllers

@inproceedings{Theobald2011CentralWR,
title={Central water resources management in a cascade of hydropower plants},
author={Ute Theobald and Stephan Theobald},
booktitle={ICNSC},
year={2011}
}