Artificial intelligence is a new ally to combat water leaks

According to the National Sanitation Information System (SNIS), 40% of water intended for consumption is lost in distribution. It would be enough to supply around 66 million people in one year, almost twice as many as the 35 million Brazilians without access to drinking water. Artificial intelligence can change this scenario.

The work to find leaks is slow and inefficient. Supply companies use a geophone, an electronic device with an acoustic system, which works as a kind of “asphalt stethoscope”, capable of “listening” to the noise of the pipes and thus detecting leaks.

In search of a more efficient technology, Marília Lara and Antônio Oliveira created the startup Stattus4, in Sorocaba (inland of SP). The first device developed was an artificial intelligence system called 4Fluid Móvel, capable of identifying noise in underground pipelines and generating reports on the area.
Unlike the geophone, which needs a technician to operate it, the device can be used by anyone. It is a collector with a rod, which, when placed against the hydrometer, captures the sounds in the pipes in ten seconds.

Having the analysis, the supply company only sends employees to the place of the problem. “The team no longer needs to travel around the city looking for leaks. The hardware guarantees that the data is accurate. Our product is twice as fast as a geophone and, on average, costs half the price”, says Marília.

The value of the service varies according to the size of the supply concessionaire, the number of inhabitants in the municipality and the number of leaks in the region.

“Today, 53 Olympic swimming pools are saved every day by our technology. The goal is to reach 1,000 a day in three years”, says Marília.

Among the customers is Grupo Águas do Brasil, which serves cities such as Niterói (RJ), Petrópolis (RJ) and Votorantim (SP). Together, the three have around 940,000 people.

In general, water losses occur along the network and distribution branches, as a result of leaks in the pipes or problems such as poor calibration of water meters, reading errors, fraud and clandestine connections.

Identifying and predicting leaks in the network was the challenge accepted by Thommas Flores, a fellow at the Laboratory of Energy and Hydraulic Efficiency in Sanitation (LENHS) at UFPB (Federal University of Paraíba).

“With AI (artificial intelligence) it is possible to take the consumption pattern of a supply system and compare it to consumption at any moment, and thus identify leaks”, says Flores, now a master in electrical engineering.

The project has already created an algorithm and is being tested in the water network of the UFPB laboratory. There are plans to apply the system to the João Pessoa (PB) network. Around 820,000 people would be impacted.