Reducing Water Leaks by Managing Smart Pressure

i2O Water Develops Smart Pressure Management Systems for Water Distribution Networks, which It Claims Can Reduce Losses Due to Leakage by an Average of 20 per cent

Water pressure and flow are monitored every few seconds at various points in the distribution network by data loggers that sit in the water pipeline, creating a continuous stream of time-series data. This data is transmitted wirelessly to a cloud-based data management platform, where it is amalgamated to build a real-time map of the entire system. Based on this information pumps and valves can be fine-tuned to optimise water pressure and flow, either by remote control or in real-time by fully automated processes.

i2O's solutions are built on what it calls a 'self-learning open-loop architecture', which is intelligent and event-driven. Events are taken up by software services, generating further events.

"The system consists of a set of loosely coupled services that intercommunicate via a message broker," says Williams.

He continues: "If one of our logging devices sends some data to our platform, an event would be when we record the data, validate the data or analyse the data. One of these events might then get picked up by our alarming service, and that would raise an alarm event; for example, high pressure. Something might then respond to that alarm event - and so on."

An important feature, says Williams, is that the system can replay historic event data. This means that when a new service is added it can 'learn' from the historic record and so be operational in the shortest possible time.

"When a new service is added to the ecosystem it requires a base of historical statistics, for example how many times we get this or that type of alarm. By replaying all the alarm events it can build up the historic stats to provide that foundation. I think that's unique," he says.

The cloud-based platform is hosted by RackSpace and consists of about 20 virtual machines. While i2O has been operational since 2005,in 2012 it began migrating away from its existing Microsoft SQL Server-based systems and on to Apache Cassandra, the NoSQL database supported by DataStax.

"It was an opportunity to re-architect our SaaS offering," says Williams.

Going into more detail, he explains that each of the software services that "add value to the system" need to store data in different ways. That might be time-series data from loggers or discrete event data. This variety makes NoSQL a more natural choice than the more rigid schema-based SQL.

The practicalities of providing visibility for customers into their water distribution systems were another consideration, including the visualisation of 'time slices' of historic data.

"A large fraction of what the i2O solution does is around gathering monitored data, learning and automatically acting upon it.

"For a typical installation we build up a huge amount of time-series data and the fast retrieval of that data is very important. We do lots of data aggregation. It is important that the user can look at the data at different time aggregation levels - we provide tools for that.

"As they zoom in they get more detail, but we don't want to consume huge amounts of bandwidth transferring all the data, so we only supply the very high resolution data when the user has zoomed in to the appropriate level.

"We realised we were going to need to look for something that was fundamentally different to a relational database; we needed a column-oriented store," Williams explains.

Cassandra was chosen over HBase and "an offering from IBM".

"Cassandra was by far the fastest. It was speed of retrieval more than speed of writing, although Cassandra is pretty quick at that too as it turns out.... Scalability was also important. The distributed nodal nature and scalability was far greater than the other offerings that we saw."

i2O makes use of the community version of Cassandra.

"We're pleased with the support we can get from the community. We don't need the additional pieces offered in the commercial version - although they're thinking we should be using Hadoop - but we have our own algorithms so don't feel there's any benefit yet for that."

Asked whether i2O's technology might be applicable in other IoT settings, Williams says he has been in discussions with other players.

"We've talked to a number of smart meter vendors and their architecture is not nearly as sophisticated as ours. When we talk about ours they suddenly realise why we've taken this route - and one of them in fact has just signed a global partnership arrangement with i2O to distribute our products," he says.

Source: Computing

Read More Related Content On This Topic - Click Here

Media

Taxonomy

• Technology
• Pipes Design