School of Mechanical Engineering Seminar
Monday, December 2, 2015 at 15:00
Wolfson Building of Mechanical Engineering, Room 206

Water quality model based only on water quality online measurements

Bar Amit

MSc Student of Prof. Lev Shemer and Dr. Eyal Brill

 Abstract

The objective of the water quality model is to predict water quality transport and its fate, throughout the various stages of water distribution systems.

A water quality model can provide many engineering insights on the characteristic behavior of water, such as the concentration of dissolved matter within the distribution system and, in particular, of a disinfectant (e.g., chlorine) in key locations within water distribution networks. It can also assist managers in performing a variety of water quality-related tasks, such as determining the water source percentage at each location, etc.

Water flow pathways as well as travel times through distribution systems are highly variable. This is due to the looped layout, the relatively large number of available pathways, as well the continuous changes in water usage over space and time. This makes modeling a complex task, which requires many inputs (both hydraulic and water quality field measurements).

The increasing numbers of online sensors may be used for water quality modeling. The different sensor types and multiple locations within the distribution network can provide insight on the nature of the water flow and the change in the dissolved content of different /substances.

There are several methods for modeling the transport of chemicals in water distribution systems. The advantages and disadvantages of those methods are discussed below.

We propose a different and novel approach by suggesting that new data made possible by the application of numerous sensors can be utilized as the only input for a water quality model. This can be done by estimating both the detention time as well as the concentration (by estimating norm change) at any location between two measuring points, without the need to use the transport and mixing equations and without the use of a hydraulic model.

The proposed solution was tested on real data from two different water distribution networks (two test environments). One has three measurement stations; the second has two measurement stations with different sensor types and different distances between the stations.  

The results were tested statistically using confidence interval limits and were found valid for both prediction of detention time and for the norm change, with the exception of norm change predictions for larger distances between the stations.

The detention time was compared to the hydraulic model and produced similar outputs.