Search

Multipurpose water systems are used to deal with multiple objectives related to the usage of water for daily human activities. These activities are often conflicted which creates a challenging water management task. To provide reliable water resources management decision support tools for successful forecasting of hydraulic data (river flows and water levels) are essential. This research presents an approach for forecasting river water levels influenced by hydropower plant operations using artificial neural networks. This approach estimates hourly water level fluctuations at the control location using the water levels and hydropower plant discharge data as input. This tool can be used for fast assessment of different hydropower plant operation plans and help in choosing the optimal one. This water level forecasting procedure is applied and tested on the Iron Gate water system, placed on the Danube River, to deal with multiple objectives in water system management (hydropower production, flood protection, and inland navigation) and shows promising results.

The objective of this research is to propose a novel framework for assessing the consequences of hazardous events on a water resources system using dynamic resilience. Two types of hazardous events were considered: a severe flood event and an earthquake. Given that one or both hazards have occurred and considering the intensity of those events, the main characteristics of flood dynamic resilience were evaluated. The framework utilizes an artificial neural network (ANN) to estimate dynamic resilience. The ANN was trained using a large, generated dataset that included a wide range of situations, from relatively mild hazards to severe ones. A case study was performed on the Pirot water system (Serbia). Dynamic resilience was derived from the developed system dynamics model alongside the hazardous models implemented. The most extreme hazard combination results in the robustness of 0.04, indicating a combination of an earthquake with a significant magnitude and a flood hydrograph with a low frequency of occurrence. In the case of moderate hazards, the system robustness has a median value of 0.2 and the rapidity median value of 162 h. The ANN’s efficacy was quantified using the average relative error metric which equals 2.14% and 1.77% for robustness and rapidity, respectively.

The paper presents a functionality investigation of the key dam elements based on finite element analysis. A detailed analysis of filtration processes, dam strength, and the surrounding rock mass was conducted. Dam elements whose potential damage could jeopardize the normal functioning of the embankment dam have been identified. A particular emphasis was placed on the analysis of dam elements that have been identified as weak points. A numerical analysis of the impact of individual grout curtain zone failure on leakage under the dam body, a strength analysis of the overflow section, as well as the analysis of the slope stability that can compromise the functioning of the spillway have been performed. To analyze the partial stability of individual structural elements, a new measure of local stability was introduced as the remaining load-bearing capacity. As a case study, the Zavoj dam, which is a part of the Pirot reservoir system in the Republic of Serbia, was used. Investigation revealed that local damage to the grout curtain will not significantly increase leakage under the dam body, the overflow section is one of the most robust elements of the dam, but the slope above the spillway can compromise the functioning of the overflow and thus the safety of the entire dam. Based on the analysis of the results of the remaining load-bearing capacity, the dependence of the spillway capacity on earthquake intensity has been defined. The established relationship represents a surrogate model for further assessment of dynamic resilience of the complex multipurpose reservoir system, within the scope of the advanced reservoir system management.

Dam and reservoir systems (DRSs) are crucial aspects of the infrastructure necessary for reliable water resource management. Nowadays, DRSs are being increasingly affected by numerous natural and anthropogenic impacts (aging and outdated infrastructure, climate change, natural hazards, global crises, etc.). Hence, additional pressure on DRS management is being applied as DRSs must be operated in adverse operating conditions, outside of their design envelopes. Since there is no practical way to redesign DRSs to meet all possible adverse conditions, efficient simulation tools are necessary for various “what-if” analyses. A system dynamics (SD) approach can be used, as it has shown the capacity to comprehend the intrinsic system complexity. In this paper, an 11-step framework for the dynamic modelling of reduced functionality in a DRS and the emulation of the system operation in adverse conditions is proposed. The framework covers the system model design, input scenario generation, system simulation, and performance evaluation steps. A focus is placed on the steps related to system decomposition, the identification of failure-indicative parameters, the definition and implementation of failure functions in the subsystem dynamic models, and dynamic failure modelling. Through these steps, a novel procedure is proposed for the dynamic modelling of the DRS subsystems’ failures (reduced functionality), common in the operation of DRSs under adverse conditions. For each subsystem prone to failure, failure-indicative parameters are identified. Failure functions employing generic functionality indicators, with values spanning from 0 to 1, are suggested to modify the values of the failure-indicative parameters in simulations and emulate the component failure impacts on DRS operation. Possibilities for modelling failure modes for different subsystems, varying in nature, duration, and magnitude are discussed. Potential physical damage to the system components, increases in measurem

To allow for the reliable hydraulic efficiency estimation of the tubular turbines before and after their revitalization at the Hydro Power Plant (HPP) Iron Gate 2, a novel flow measurement system was designed and tested on two turbines. In this paper the details of used approach for the assessment of the flow measurement uncertainty are presented. The presented methodology is tailored for specific application of the Velocity – Area flow measurements at HPP, based on the utilization of electro-magnetic velocity meters (EMVM), instead of traditional propeller current meters. Although the procedure for the assessment of the flow measurement uncertainty in similar cases of low-head plants and short intakes can be found in the literature, the adverse flow conditions, novel EMVM sensors and the specific operating modes have instigated a need for modifications and improvements of the procedure. Assuming analogy between the current meters and EM meters, the IEC 60041 and ISO 3354 procedure for flow rate measurement uncertainty assessment was used as a template, with implemented moderate changes accounting for the features of the applied measurement. Selected flow measurements from the 2020. campaign are analyzed, with an emphasis on the variation of the magnitudes of the measurement uncertainty components.

Flow measurement of low head turbines, with short intakes like bulb turbines is challenging since there is no “regular” cross-section with fully developed velocity profile. In most situations the flow data during field turbine acceptance tests are obtained using the index method, with flow coefficients transferred from physical model tests done in the laboratory. In the non-standard situations, with adverse flow conditions this may lead to unpredicted flow rate uncertainty. This paper presents the used methodology and results of flow measurement at inlets of two bulb turbines of Hydro Power Plant (HPP) Iron Gate 2 (Danube river). Turbines are Kaplan, with low head (2.5 – 12.5 m), 27 MW and maximal flow rate 420 m3/s. Due to HPP’s disposition the inflow angle (in a “horizontal” plane and in respect to the turbine axis) can be up to 400. Even reverse flow directions in upper regions of cross-section can occur. The movable 14.5x3.1 m steel frame, shaped to minimize flow disturbances was used at the intake of turbine, upstream of the trash rack, to traverse the inflow cross section 14.5x26 m. The 15 spherical 3D electromagnetic velocity meters capable of bi-directional measurements were mounted on the frame, while redundant measurement were made using 2 ADVs. Two measurement strategies were used: incremental with 18 profiles and 10 min averaging time at each profile and continuous using constant lifting speed of 0.05 m/s. Uncertainty assessment procedure yielded discharge measurement uncertainties over 1 % for incremental, and up to 5 % for continuous traversing.

This paper provides insight into the Institute's long-standing engagement in the design and implementation of hydroinformatics systems in Serbia and the region. The introduction provides a definition of the hydroinformatics system and different areas of application, as well as an overview of the best practices in the world. An overview of the general platform that was created based on experience in the development of different systems has been presented here. We have described the functionalities integrated into a single software platform based on mathematical models and computational services. Various practical examples of application by the Institute are presented with the specifics of implementation in line with the purpose and characteristics of the studied systems. The conclusion highlights the role of applied hydroinformatics systems and the effects of application by users. Possible further development and implementation directions in water management and hydropower systems in Serbia and the region have been also presented.

Machine learning methods have been widely and successfully applied in hydrological problems. Most of the methods, such as artificial neural networks, have been focused on estimating hydrological data based on observation over time. Even though these models provide good results, it can be observed that results become unreliable when the training dataset is small or when input data is significantly out of range compared to the training data. Therefore, a new approach is presented, in which artificial neural networks are trained to satisfy physical laws. This is conducted by a novel method called physics-informed neural networks (PINNs), in which physical principles are embedded in a custom loss function. This paper presents the application of physics informed neural networks for solving 1D flood wave propagation in open channels. The research has shown promising results.

Reliable water resources management requires decision support tools to successfully forecast hydraulic data (stage and flow hydrographs). Even though data-driven methods are nowadays trendy to apply, they still fail to provide reliable forecasts during extreme periods due to a lack of training data. Therefore, model-driven forecasting is still needed. However, the model-driven forecasting approach is affected by numerous uncertainties in initial and boundary conditions. To improve the real-time model's operation, it can be regularly updated using measured data in the data assimilation (DA) procedure. Widely used DA techniques are computationally expensive, which reduce their real-time applications. Previous research shows that tailor-made, time-efficient DA methods based on the control theory could be used instead. This paper presents further insights into the control theory-based DA for 1D hydraulic models. This method uses Proportional–Integrative–Derivative (PID) controllers to assimilate computed water levels and observed data. This paper describes the two-stage PID controllers’ tuning procedure. Multi-objective optimization by Nondominated Sorting Genetic Algorithm II (NSGA-II) was used to determine optimal parameters for PID controllers. The proposed tuning procedure is tested on a hydraulic model used as a decision support tool for the transboundary Iron Gate 1 hydropower system on the Danube River, showing that the average discrepancy between modeled and observed water levels can be less than 0.05 m for more than 97% of assimilation window.

The objective of this research is to introduce a novel framework to quantify the risk of the reservoir system outside the design envelope, taking into account the risks related to flood-protection and hydro-energy generation under unfavourable reservoir element conditions (system element failures) and hazardous situations within the environment (flood event). To analyze water system behavior in adverse conditions, a system analysis approach is used, which is founded upon the system dynamics model with a causal loop. The capability of the system in performing the intended functionality can be quantified using the traditional static measures like reliability, resilience and vulnerability, or dynamic resilience. In this paper, a novel method for the assessment of a multi-parameter dynamic resilience is introduced. The multi-parameter dynamic resilience envelops the hydropower and flood-protection resilience, as two opposing demands in the reservoir operation regime. A case study of a Pirot reservoir, in the Republic of Serbia, is used. To estimate the multi -parameter dynamic resilience of the Pirot reservoir system, a hydrological model, and a system dynamic simulation model with an inner control loop, is developed. The inner control loop provides the relation between the hydropower generation and flood-protection. The hydrological model is calibrated and generated climate inputs are used to simulate the long-term flow sequences. The most severe flood event period is extracted to be used as the input for the system dynamics simulations. The system performance for five different scenarios with various multi failure events (e.g., generator failure, segment gate failure on the spillway, leakage from reservoir and water supply tunnel failure due to earthquake) are presented using the novel concept of the explicit modeling of the component failures through element functionality indicators. Based on the outputs from the system dynamics model, system performance is determine

Increasing renewable energy usage puts extra pressure on decision-making in river hydropower systems. Decision support tools are used for near-future forecasting of the water available. Model-driven forecasting used for river state estimation often provides bad results due to numerous uncertainties. False inflows and poor initialization are some of the uncertainty sources. To overcome this, standard data assimilation (DA) techniques (e.g., ensemble Kalman filter) are used, which are not always applicable in real systems. This paper presents further insight into the novel, tailor-made model update algorithm based on control theory. According to water-level measurements over the system, the model is controlled and continuously updated using proportional–integrative–derivative (PID) controller(s). Implementation of the PID controllers requires the controllers’ parameters estimation (tuning). This research deals with this task by presenting sequential, multi-metric procedure, applicable for controllers’ initial tuning. The proposed tuning method is tested on the Iron Gate hydropower system in Serbia, showing satisfying results.

Model-driven forecasting used for flood risk assessment or river hydropower systems management, can produce bad results due to many model uncertainties. False inflow and lateral inflow data and/or poor estimation of initial conditions are some of the uncertainty sources. To improve model-driven forecasting, data assimilation methods are used for updating model (e.g., water levels) according to measurements. Widespread data assimilation methods (EnKF, Particle Filter) often increase computational time, which creates difficulties in everyday application of these methods in hydraulic modelling. This paper presents novel approach based on indirect model update adding correction flows at observation locations. This novel, tailor-made, assimilation approach uses proportional-integrative-derivative controller’s theory as algorithm for correction flow calculation. Using indirect approach for model update has justification in models where multiple inflows, including lateral inflows, are bad estimated or even neglected. This novel approach is tested on 170km long section of Danube model in Serbia, showing good performance.

This paper investigates how to transform a manual monitoring system into a smart environment using IoT technologies. Large dams are of huge importance as water reservoirs and potential energy sources primarily because of their use for electric energy generation. Monitoring and predicting dams' behavior are quite challenging. Due to a variety of limitations, environmental obstacles and dam system complexity, the application of RFID, Bluetooth, and IoT technologies has proven to be an adequate, affordable and reliable dam safety solution. This paper introduces a comprehensive model of dam safety management and monitoring based on pervasive technologies. The model described in this work has been applied within the large dam "Iron Gate 1", on the river Danube in the Republic of Serbia. The results show that the use of new technologies in the dam monitoring process can reduce human error and improve overall process.

This paper discusses harnessing Internet of Things in monitoring and managing dams in Republic of Serbia. Large dams are of major importance, primarily because of their use for electricity, but risks which are associated with it should be greatly taken into account. There is a need to consolidate information related to dam facilities in order to use them for dam management in the Republic of Serbia. An information system has been developed based on the existing systems, allowing utilization of intelligent network sensors. The aim of the paper is to describe possibilities of the Internet of Things application within a specific system for dam safety management. In order to facilitate the inclusion of a large number of intelligent sensors, a new data acquisition module for communication with sensors in the monitoring network is defined. The system should provide on time alerting in case security parameters deviate from the expected values.

Indikatori ponašanja brane su relevantne veličine, čijim se praćenjem utvrđuje da li je stvarno stanje brane u eksploataciji u saglasnosti sa onim što je predviđeno i očekivano u fazi projektovanja. Veličine koje se prate treba da se kreću u nekom unapred definisanom opsegu koji garantuje stanje stabilnosti brane. U ovoj disertaciji su predloženi različiti pristupi modeliranja i parametarske identifikacije indikatora ponašanja brane, poput horizontalnih pomeranja i nivoa vode u pijezometrima, tehnikama računarske inteligencije. Prvi pristup je da se linearno preslikavanje uzročnih veličina u indikatore ponašanja, koje se koristi kod višestruke linearne regresije, zameni nelinearnim. Drugi pristup, predložen u ovom radu, zasniva se na primeni postupka parametarske identifikacije nelinearnih sistema. Horizontalna pomeranja i nivoi vode u pijezometrima su nelinearne, složene funkcije uzročnih veličina, pa je za njihovo modeliranje korišćena NARX (Nonlinear Auto Regresive eXogenous- nelinearni auto-regresioni model sa spoljašnjim ulazom) struktura, kojom je opisana široka klasa nelinearnih dinamičkih procesa. Predloženi pristupi formiranja modela primenjeni su za modeliranje i parametarsku identifikaciju horizontalnih pomeranja tačaka brane Bočac, kao i nivoa vode u pijezometrima brana Đerdap II i Prvonek. Nelinearni modeli zasnovani na tehnikama računarske inteligencije implementirani su korišćenjem programskog jezika Java i programskog paketa Matlab. Tehnike računarske inteligencije korišćene u ovom radu su višeslojni perceptron, RBF (RBF - Radial Basis Function – radijalna osnovna funkcija) neuronska mreža i ANFIS (ANFIS - Adaptive-Network-Based Fuzzy Inference System - fazi sistem za zaključivanje zasnovan na adaptivnoj mreži). Nedostajući podaci u skupu merenja mogu biti uzrok problema u okviru procesa učenja i loših performansi dobijenih modela. U cilju nadomeštanja nedostajućih podataka korišćene su tehnike iz domena matematičke statistike. Prisustvo autlajera u, The dam behavior indicators are relevant factors whose monitoring indicates whether the actual operational state of the dam is in accordance with what is expected and anticipated in the design phase. Such indicators should move in a predefined range, in order to guarantee stability of the dam. This dissertation proposes different approaches to modeling and parametric identification of the dam behavior indicators, such as radial displacements or piezometric water levels, using the techniques of artificial intelligence. The first approach is to replace linear mapping of causal variables into behavior indicators, which is used in multiple linear regression, with nonlinear. The second approach proposed in this paper is based on applying the method of parametric nonlinear system identification. Radial displacements and piezometric water levels are nonlinear, complex functions of causal variables, so for their modeling NARX (Nonlinear Auto Regresive eXogenous), which is employed to describe a wide class of nonlinear dynamic systems, is used. These proposed approaches are used for modeling and parametric identification of radial displacements of dam Bočac, and piezometric water levels of dams Iron Gate II and Prvonek. Nonlinear models based on artificial intelligence techniques have been implemented using the Java programming language and MATLAB. Artificial intelligence techniques used in this work are the multilayer perceptron, RBF (Radial Basis Function) neural network and ANFIS (Adaptive-Network-Based Fuzzy Inference System). The presence of missing data in a set of measurements may be causing problems in the learning process and the poor performance of the obtained models. In order to predict the missing data, the techniques of mathematical statistics have been used. Outliers present in a set of measurements have a big effect on the prediction of missing data, and their presence is specifically analyzed. The problem of optimizing the inputoutput model, which involves d