Data Science Investment Counter —Funding raised by UK data science companies in 2017 as of today
£ 342.935 Million

Can Data Science Solve Leaky Pipes?

When it comes to maintenance, water networks are a hard nut to crack.

Water companies often struggle to understand what parts of the network should be fixed, and what pipes are at risk of bursting and need to be substituted.

Some companies try to solve that by just setting aside large amounts of resources every year, to invest into indiscriminate pipe replacement— but that is inefficient and financially onerous.

A spin-off  company of Imperial College’s InfraSense Labs is trying to solve this with data science and innovative instruments.

London-based company Inflowmatix has developed a high-frequency pressure analysis device that can be plugged into any part of a water network and detect  changes in pressure or performance—which could be the harbingers of leaks, bursts and malfunctions.

inflowsense

Inflowmatix’s device (pictured) can assess the health of a network by gleaning up to 128 measurements of water pressure every second. Pressure-gauging devices currently in use generally gather similar data only once a minute.

Company’s CTO Mike Williams says the device can assess the health of a network by gleaning up to 128 measurements of water pressure every second. Pressure-gauging devices currently in use generally gather similar data only once a minute. 

“So we are able to generate a large amount of additional  data which we can analyse to look for interesting patterns in the network, and spot possible transients [i.e. glitches or unexpected variations],” Williams says.

“We can also compress the data and transmit it into our cloud servers for algorithmic processing. Our intent, currently, is to move those algorithms downstream, to have them run on the device itself: the nearer to the source of the data we can perform the analysis, the quicker the value customers can derive.”

The technology, Williams says, will allow companies  “to redirect millions to areas where pipes are more likely to rupture, which would save money spent on the ones they repair but aren’t going to burst.”

The company is currently deploying about 200 devices in a test project involving three companies managing water supply systems (Inflowmatix’s technology is not applicable to wastewater networks.)

Right now, Inflowmatix’s device and algorithms are only able to spot whether a pipe, or a larger portion of the network, are showing irregularities that could end up causing problems. But the company plans to beef up its technology to make it able to return more specific recommendations.

“We’re trying to investigate how transients could cause bursts—to understand  how these transients affect the probability of pipes bursting,” says Inflowmatix’s data scientist Vasilis Kontis. “In this was we could help water companies reduce these incidents by quickly optimising their water networks.”

“That involves statistical methods, models to estimate the effect of pressure on bursts and, further down the line – when we obtain enough  data— we can streamline these process and update our model estimates automatically.”

Featured image via bit.ly/2fKXSfQ

SHARE THIS ARTICLE:
BY SHACK15

Co-working space and blog dedicated to all things data science.

Subscribe to our newsletter