Leak localization in complex water pipeline networks using Multimodal Sampling

Pipeline networks are used commonly for transporting water, gas, and oil in industrial as well as residential areas. Pipelines can have leaks due to various reasons such as blockage, erosion, overflow of fluid and in the wake of natural disasters, and this cost billions of Saudi Riyals every year due to wastage of water, and the damage it causes to the surrounding infrastructure. Hence, any engineering solution that efficiently addresses this vital issue will prove to be of immeasurable economic benefit. Monitoring of complex water-pipeline networks to detect and locate leaks remains a challenging task. A network of Wireless Sensors is very effective for collecting data, since they do not require messy wiring, unlike the wired ones, and the data collected can be communicated effectively. However, since the wireless sensors have limited energy, the challenge is to use the energy conservatively, and at the same time, monitor the pipeline accurately for leaks. One effective approach to tackle this challenge is to use hierarchical (also called multiscale) sampling which is based on the information data from various sensors being fused together to get better results, and on these sensors which could have different fidelities and levels of energy consumption, being either sparsely or densely distributed in the network, as and when deemed necessary. The main aim of this research is to extend the work carried out in our past projects for detecting and locating single leaks in a straight pipeline to doing so in a complex pipeline network which may include several bends. It is proposed here that this can be done effectively by fusing pressure and vibration measurements from various locations in the network, the former be used for detecting the leak and getting closer to its location, and the latter for pinpointing the exact location of the leak. This research will also address the problem of the optimal locations in the complex network of these two types of sensors, and the development of a scheme to fuse the information from these sensors to detect and locate a single leak. The different methods that exist in the literature for modeling pressure and vibration sensors, and which are used in complex pipeline networks will be studied here, and a new multimodal scheme for leak localization that is suitable for Wireless Sensor Network (WSN), will be developed and evaluated using EPANET and MATLAB simulation. Our group has already several publications in the area of pipeline monitoring and leak detection. We expect that the achievements of this project will result in a conference and a journal paper.