Castro Gama M.,UNESCOIHE Institute for Water Education |
Popescu I.,UNESCOIHE Institute for Water Education |
Popescu I.,Polytechnic University of Timişoara |
Mynett A.,UNESCOIHE Institute for Water Education |
And 2 more authors.
Transactions of the ASABE | Year: 2016
Severe flood events in China are a common cause of life losses. Many efforts have been carried out by the Yellow River Conservancy Commission (YRCC) to understand flood development and impact on the Yellow River. Modeling approaches based on discretization of the modeled domain in square and rectangular grids have great importance in the management of rivers, but they usually present two drawbacks: the required accuracy of the meandering of wide long rivers is not well represented, and the need to use many grid cells reduces computational speed. Modeling and simulation of extreme flood events using unstructured grids has not yet been applied to the Yellow River. This study shows how such a method can be beneficial to decision makers in the Yellow River region in case of flood events. The research shows that a flexible mesh discretization of the domain can overcome the two drawbacks. Along with a test of the proposed modeling method, new characteristics of the spatial evolution of flood events in the Yellow River emerged and are presented, showing the capabilities of DFloridaOW-FM software in modeling such a complex system. © 2016 American Society of Agricultural and Biological Engineers.
Dhakal N.,UNESCOIHE Institute for Water Education |
Salinas Rodriguez S.G.,UNESCOIHE Institute for Water Education |
Schippers J.C.,UNESCOIHE Institute for Water Education |
Kennedy M.D.,UNESCOIHE Institute for Water Education |
Kennedy M.D.,Technical University of Delft
Desalination and Water Treatment | Year: 2015
In this research work, the induction time of two brackish water reverse osmosis (BWRO) plants was measured with and without antiscalant (AS) operation with focus on calcium carbonate precipitation. This study focused on two BWRO plants in the Netherlands. The scaling potential of RO concentrate in the plant at recovery of 80% was calculated using the PHREEQC program. Induction times in RO concentrates were measured by collecting RO concentrates directly from the plants, with and without an AS dose in an air-tight glass reactor. The solution was continuously stirred for homogenization and maintained at room temperature using a thermostat. The change in pH of the solution over time was monitored. The time required to change the nucleation phase to just the start of the crystal growth was noted from the pH versus time graph, which was defined as induction time. At the end of the experiments, precipitate with RO concentrate with and without AS dose (plant B) were collected and an X-ray powder diffraction (XRD) and scanning electron microscope (SEM) analysis were performed. The results showed that the measured induction times for RO concentrate with AS dose in the two plants were longer than 100 and 280 h, respectively. The plants were operated at the same recovery (80%) but with different types of feed water and different types of AS. In a plant where the induction time could be measured without AS, the induction time turned out to be short namely about 4 h. The XRD results of this plant with and without an AS dose revealed calcite as the exclusive precipitate. The shapes of the crystals obtained with AS, shown by SEM images, were smaller, less clustered, and had more rounded edges than crystals without AS. The results suggested that the measured induction times of 100–280 h in both plants are much higher since the detention time in RO systems is just 1–1.5 min and even seconds in the last membrane. Therefore, for safe operation, induction time lower than what we measured in two BWRO plants might be needed. © 2014, © 2014 Balaban Desalination Publications. All rights reserved.
Siek M.,UNESCOIHE Institute for Water Education |
Solomatine D.,UNESCO-IHE Institute for Water Education |
Solomatine D.,Technical University of Delft
Proceedings of the International Joint Conference on Neural Networks | Year: 2010
This paper describes a new forecasting technique based on multi-model ensemble in high-dimensional chaotic system. A chaotic model is built from the time series reconstruction in the time-delayed high-dimensional phase space. The chaotic model forecasts are made by the adaptive multi-local models constructed based on the dynamical neighbors found in this space. We utilize several different predictive local models (including MLP ANN) and ensemble their model forecasts to create a more accurate hybrid model. This proposed method was implemented and tested for building a storm surge model for the North Sea. The model results showed that the multi-model ensemble model has a significant increase on the forecasting accuracy compared to standard chaotic model or global neural network model. © 2010 IEEE.