Big Data and Water Supply

Between 60-70 percent of our bodies is made up of it. 370 liters of it is consumed per person in France and there is about 1.26 quadrillion liters of it on Earth. So what is it? Water of course! Water is one of the most basic elements on Earth and very often considered in surplus. But did you know that about 14 million US dollars of water is not billed every year? In developed countries there lost water accounts for 20 percent of the total water usage. Naturally, this would seem like a real world problem- completely independent from the internet. But water networks are cosistently monitored by sensors and this data is then relayed back to computers. And just as a sensor can turn off heat for a short period of time to save electricity, this same process can be applied to water.

Non-revenue water is referred to as water that does not reach the customer for some reason after it is produced thus making it lost. This lost water can result from physical losses such as leaks or from theft and other inefficiencies, also called apparent losses. Regardless of the type of loss, however, non-revenue water is damaging to the quality of water actually received by customers as well as the profitability of water utilities.

With the emergence of big data, however, there is a huge potential for this issue facing the water industry to be remedied because of three key factors. First, using big data allows us to analyze all the data that's available. In the past, we would only take a small subset and use that as the base for the entire network. But no longer. No, now we have the potential to look at all the data, which increases the potential four us to find an inefficient use of water. Additionally, the advent of complex algorithms allows us to crunch this data and discover new insights we couldn't have fathomed in the past. Finally we're now able to analyze these results in real time. Before the emergence of big data it would be six to twelve hours before the data could possibly start to be analyzed. Now, however we receive an analyze the data as it occurs representing huge potential water savings

If pursued, both internal data- like sensor and network operations data- and external data such as weather conditions should be accounted for, similar to the way CaptainDash treats inputs. Using this information, one could generate a typical “behavior” for the network in a particular area such as a few blocks or small village.

By comparing two regions that are similar in demographics, size, and income, they can detect happenings of inefficient water usage as they occur in different areas. If there is a sudden increase in the amount of water used in one area, this will indicate a possible issue. This, in turn, will cause an algorithm to begin searching for the root of the problem. However, if there is a large increase in both of the regions, that could indicate an uncontrollable factor. For example, if its a warm day, there is going to be a natural increase in water usage.

As always, The Captain