How AI and data turn city water management from an art to a science

In the face of futuristic urban change, water has been described as one of the least-disrupted municipal systems, despite its importance for survival.

The vast majority of water infrastructure in the United States is at least 50 years old, and with that age comes issues including leaks and poor water quality among others. The American Society of Civil Engineers (ASCE) estimates there are 240,000 water main breaks nationwide each year, while a News21 investigation in 2017 found that 63 million people have been exposed to unsafe drinking water. These troubling statistics are due in part to the deterioration of water infrastructure.

Some cities, like Akron, OH and Washington, DC, have turned to drones, robots and automated systems to move water management into a new era — but this is not the norm. Jeff Bronowski, Akron's water bureau manager, said more people in the water sector need to be made aware of the benefits of new technologies.

"I think the industry as a whole, there needs to be a lot more awareness as to the value of these technologies and the higher quality water that can be provided to customers when you have all this information available to you," Bronowski told Smart Cities Dive.

The state of water systems

Like much of America's core infrastructure, current water systems have received a failing grade from industry experts. In its 2017 report card on the sector, the ASCE graded drinking water infrastructure a "D" and wastewater a "D+." The group said there are around 240,000 water main breaks a year, wasting over 2 trillion gallons of treated drinking water.

The American Water Works Association believes an estimated $1 trillion is needed to maintain and expand service to meet water demand for the next 25 years.

Kim Nelson

Executive director, state and local government solutions, Microsoft

Digital technology could help ease that backlog of maintenance and move the sector forward, according to a white paper released in June by the International Water Association (IWA) and water technology company Xylem. The report urges leaders in the water sector to think boldly about how they can use digital tools, and the impact it could have.

That includes calling on water companies to create an "innovation culture" that encourages staff at all levels to find technological solutions; build consensus and a business strategy for how technology will be implemented; and develop a way to manage data that is collected in the field and use it to optimize business processes.

Katharine Cross, strategic programs manager at the IWA and a co-author of the report, told Smart Cities Dive that while the water sector has a reputation for conservatism — due to needing to minimize the risk of damaging water supplies — using technology could be key to addressing sustainability, resilience, and even behavioral and cultural issues.

"There's the potential to use these technologies to show immediate results, so for example using sensors that can detect leaks and being able to address these issues more immediately can reduce non-revenue water, so that's an immediate result," Cross said.

Report co-author Randolf Webb, Xylem’s director of global branding and partnerships, said that many solutions are already available for use, so it is imperative for executives at the top of water organizations to recognize that changes may need to be made.

"It's more at the culture level on the setting the ambition at the CEO and board level and building this mindset of needing to deploy new technologies to avert some of the challenges that continue to build," he said.

While the federal Clean Water Act authorizes federal help for water infrastructure and sewage system improvements, like much of American infrastructure, tight budgets at the local level have meant a lack of investment. Meanwhile, the federal government's previous strategy of distributing grants for improvements has been largely replaced by a focus on providing state loans that can then be sent to localities. The Congressional Budget Office (CBO) found that federal spending on transportation and water infrastructure totaled $441 billion in 2017, but that is a drop in the bucket compared to what is needed.

Cities and communities have in recent years dealt with health crises brought on by poor water quality, with more potentially to follow. The most high-profile incident came in Flint, MI, where lead infiltrated the city's drinking supply due to insufficient water treatment and led to more than 100,000 residents being exposed to high lead levels in their water. Twelve people died as a result of the contamination, many local officials were forced to resign and it prompted a federal response to try and reduce the number of lead service lines.

Meanwhile, the New York Times reported that 1,000 water districts in California are strained by underinvestment that has caused dirty water to come out of faucets and make some of the state's poorest residents sick. With those kinds of consequences on the horizon, ASCE and others have called for greater investment in water infrastructure.

Leaks and maintenance

Few things are more frustrating for city residents to deal with than water main breaks, which often result in closed and congested streets for several hours and wasted water.

Preventive maintenance, where private and city-owned water companies inspect pipes for these issues, can cause similar disruptions. To conduct inspections, water companies often need to close a road and drop a CCTV camera into a manhole to capture images. Those are are then transmitted back to a control center, where a trained inspector examines the pipes for potential issues.

Washington, DC has taken things in a new direction, reducing the need to close roads and analyze video screens. Instead, the DC Water & Sewer Authority uses autonomous robots and drones to collect video and feed it to Pipe Sleuth, a defect detection and classification software.

That software, which uses deep learning and AI, identifies and tags defects in 10% of the time of a video length — so it can process, tag and assess conditions in a 30-minute video in three minutes — and then produces a report for DC water officials to see if any areas need further attention. It also represents a significant cost saving: CCTV vehicles typically cost around $250,000 to purchase, while a robot costs $60,000.

The new system helps preventive maintenance of pipes and can also be helpful in monitoring water quality for odor and color issues, which are the main source of complaints to the city utility.

"If we can avoid a break, that's not only avoiding the customer service problem and an interruption in service, but it also reduces the cost of the operation substantially," Thomas Kuczynski, vice president of IT at the DC Water & Sewer Authority, told Smart Cities Dive. "The water quality side allows us to better plan activities to ensure the quality of the product and the age of the product in the pipe is actually improved so we can drive down the number of water quality complaints."

Sensors, drone technology and machine learning could benefit many cities and their water supplies, especially as they battle with the need to keep up consistent service. It also reduces the need for emergency fixes.

"If you use sensors and can detect that potential fracture before it ever occurs, it's cheaper to fix, you've not wasted a lot of water, you've not disrupted a city and you've taken care of a problem before it actually occurs," Kim Nelson, executive director for state and local government solutions at Microsoft, told Smart Cities Dive. "That's the power of the data, because you can analyze the data and do the predictive analysis around it, combined with things like sensors."

Customers and employees

It is not just maintenance that is being impacted by AI and other technologies. City workers' jobs are made more efficient due to modernization, while customers can reap the benefits.

In Akron, the city is looking to have a system of smart meters placed in homes by the end of 2021 that not only will collect readings on a continuous basis for billing purposes, but also will have better detection for incidents like leaks or anything else untoward.

Jeff Bronowski

Water bureau manager, City of Akron, OH

"We're going to know if there's just things that may be unusual with the customer's use that we can notify them on, like we think you have a leaking toilet, or we think you may have a burst pipe, or we think that someone might have left a faucet or spigot on inadvertently," Bronowski said. "Very soon, Akron's going to have the capability to provide that customer service that we just don't have the capabilities to provide now."

A big point of emphasis for DC's water authority is cutting its energy costs by using AI and machine learning, with those cost savings then hopefully being passed on to its customers. Kuczynski noted that the city’s Blue Plains facility is the largest advanced wastewater treatment plant in the world, so any savings for customers that can be driven through optimizing its use are ideal.

"We are the largest consumer of electricity in DC, so to the extent that we can build a model that optimizes the performance of our pumps and things, then we're able to drive down the cost of energy, which is another direct benefit for the customer," he said.

Meanwhile, for employees of city water companies, jobs are becoming automated as part of an overall trend of labor-intensive positions being more reliant on technology and AI. Bronowski said that means that operator positions are "more emergency responders and less operators," with the city working hard to update job descriptions to consider the new technology available that does the bulk of the labor automatically.

Typically, operators used to be what Bronowski described as a "labor-type mechanic position," whereas now they are primarily tasked with ensuring everything works properly and responding to issues when they come up.

"It really has turned operations from being less of an art to more of a science, because historically in the industry you had 30-year type operators that have operated these plants for a very long time and did it based on feel, based on experience," he said. "They were what I would describe as artists. Now, we're turning it into a science-based approach that utilizes a host of different variables that the human brain could never comprehend."