AI could help the water industry curb its thirst for energy

by NEW YORK DIGITAL NEWS



When Gaurav Gupta sought to make a Coca-Cola bottling plant in Abu Dhabi more efficient, AI came to the rescue.

“We were looking for a system or a tool which can help us monitor the day-in, day-out operation of the plant, help us to optimize our performance,” recalls manufacturing lead Gupta, who is responsible for production and maintenance.

Early this year, the plant began working with a Canadian startup called Pani Energy, whose cloud-based artificial intelligence software helps water treatment facilities quickly boost their efficiency and save money. Gupta and his team now receive data on the plant’s operations in real time. “If any problem comes, we are getting the warnings and alarms,” he says. As a result, the team can make faster decisions.

So far with Pani, the plant has improved its energy use ratio (EUR)—the energy required to make a liter of beverage—by 3%. Its water use ratio (WUR), or water usage compared to beverage produced, has seen the same improvement. The plant also avoided damage to a water filtration membrane that would have cost $10,000 to replace, Gupta notes.

Next up: boosting WUR to 5%. “This will help us to improve our energy use ratio,” Gupta says. In 2024, the beverage company’s wastewater treatment plant will start using Pani too. “All the water processes, we are connecting to Pani so that we will have visibility.”

Decarbonizing the water industry is the overarching goal of Pani Energy, whose clients include industrial, municipal and desalinization plants, mostly in the Middle East and the United States. For 2032, the A.I.-powered company has set an ambitious target: reducing the sector’s annual greenhouse gas emissions by 510 megatons—1% of the global total.

CEO Devesh Bharadwaj cofounded Pani in 2017, when he was still an undergrad at the University of Victoria. Bharadwaj, who grew up in New Delhi, had started off studying biomedical engineering, hoping to make a difference by building medical devices.

But after becoming fascinated by the idea of using technology to help solve humanity’s biggest problems, he concluded that climate change, energy, and water topped the list. “I noticed that there was not enough, relatively, innovation and transformation in the water sector,” says Bharadwaj, 29. “So I decided to focus my time on that and climate.”

To that end, Bharadwaj switched to mechanical engineering and launched Pani, which pivoted twice to arrive at its current business model.

First, the company developed a large-scale battery system that used desalinization to store energy. Determining that wouldn’t make a fast enough impact, Pani began using hardware and software to retrofit desalinization plants, making them more energy-efficient. Again, Bharadwaj found that approach too slow. “I realized that I would never be able do that fast enough to reduce 1% of global emissions,” he says. “So we pivoted to only a software company.”

Although the water industry might not look like a big carbon polluter, water use, storage and distribution account for a whopping 10% of all greenhouse gas emissions, according to one estimate. “Over the next 20, 30 years, it’s only getting worse,” Bharadwaj warns.

With climate change reducing access to fresh water, cities and industries are turning to desalination and reusing wastewater. In some cases, they must spend 20 times more in costs and carbon emissions to get the same amount of water, according to Bharadwaj. “Extremely intensive water treatment is now needed, which is only making the emissions profile higher, which is, again, making climate change even worse.”

Water treatment processes such as pumping, adding chemicals, cleaning devices, and disinfecting with ultraviolet light all require energy, says Madjid Mohseni, a professor in the department of chemical and biological engineering at the University of British Columbia. “No matter what type of process you look into, there is energy consumption.” With each person using an average of 500 liters of water a day, the energy demand is huge, Mohseni adds.

Reckoning that it could take 100 years to transform the water treatment sector, Pani set out to get in front of the problem by doing it in 10, Bharadwaj relates: “How can we accelerate the transformation that reduces their greenhouse gas emissions and cost profile?”

For many players, the economic argument for cutting energy use is compelling. “A lot of these municipalities and utilities are trying to reduce their energy needs, because that’s money at the end of the day,” Mohseni says. “There is a lot of push and demand for energy savings.”

Still, change has been slow, Mohseni observes. That’s partly because the water industry is risk-averse when it comes to novel approaches, given that operators are responsible for public health. For example, if something goes wrong and it leads to a shutdown or lower-quality water, operators face a public backlash. “So they’re a little bit, in general, lagging behind other sectors in terms of adopting new strategies, including AI.”

Pani, which recently made the Global Cleantech 100 list for the second year in a row, aims to change that by letting customers start wherever they’re ready. Clients, who pay an annual subscription, can “transform with the software over time,” Bharadwaj says.

While many other industries have become highly automated, water treatment still relies heavily on human decisions, he explains. Pani’s software assumes much of that burden by taking data from control systems and using it to understand a plant’s design in minute detail. In the cloud, it then simulates how the facility could be running better and makes predictions.

The software sends all that information back to the operations staff, Bharadwaj says. “In real time, just like a coach, it’s watching, analyzing all the data, and then giving analytics and insights through a web-based platform.”

For mega-scale plants, the cost savings from using Pani are 5% to 7%, which adds up to millions of dollars of a year, Bharadwaj says. Savings for smaller operations can be 20% to 30%.

Applying A.I. to the chemical engineering and biological processes used in water treatment is no easy task, Bharadwaj admits: “You have to combine standard machine learning capabilities and modeling approaches with really fundamental, deterministic physics and chemistry modeling together to be able to automate artificial intelligence.”

That means people play a key role. Besides software developers, the Pani team—which has grown to about 40 from a handful of early employees—includes chemical engineering modelers, data scientists, and process operations experts.

At the treatment plants run by Pani client H2O Innovation, equipment failures can be expensive. When a valve recently shut off at one of the water filtration giant’s industrial clients, it ruined all of the facility’s filtration membranes, recalls Paul Bartlett, director of automation and service. The tab: upwards of $30,000.

Then H2O started using Pani’s software. When the same valve shut down again, Pani instantly spotted the problem and used its automated alarm system to alert Bartlett and a colleague by text and email. “That was pretty eye-opening to me and obviously the customer as well, that within days of implementation, we were able to see a significant savings,” Bartlett says.

H2O builds, engineers, and operates large municipal as well as industrial water treatment systems. The company, which runs and maintains more than 700 facilities throughout North America, now uses Pani’s software at roughly 10 of its sites. Over the next three to six months, it hopes to grow that number to about 50, Bartlett explains.

H2O had its own data-logging software, Bartlett says, but the process engineering team would spend hours analyzing the information. Now Pani’s A.I. does much of that work.

Thanks to the software, H2O is also doing something new, he adds. “It’s allowing us, for the first time ever in our company, to monitor the carbon emissions at these plants based on chemical usage, on power consumption, things like that.”

Pani is letting the company establish a baseline for carbon use across all of its plants. “You can’t fix or change what you don’t measure,” Bartlett says. “After we monitor and establish that baseline, we’re going to be able to set goals and targets to reduce that.”

Asked how Pani plans to reach its 2032 decarbonization goal, Bharadwaj says the company now works with more than 100 facilities in roughly 10 countries. By 2032, it expects to be in 1,000 to 1,500 plants. “That’s the range we’ll need to be in, or more, to reduce about half a gigaton, 510 million tons a year.”

Pani has published a white paper that includes a roadmap for getting there. The plan has four tenets. The first: Partner with facility owners, operators and solution providers. The second is rapid deployment with low barriers to adoption—“a low-touch, low-risk model that allows facilities to quickly get started and provide strong returns on investment,” Bharadwaj says.

The third tenet is continuously investing in operational AI. That includes automating operations intelligence at facilities, Bharadwaj says, “and later helping them retrofit and actually make larger capital decisions which further reduce their cost and greenhouse gas profile.”

Pani’s fourth tenet: looking at water treatment differently, by working with large incumbents to define a new category of facility called “water zero” that views sustainability through an environmental as well as a business lens. That involves a four-step method: measuring what matters most, optimizing those metrics, deploying A.I. to accelerate retrofits, and using offsets and financing to make retrofits more attractive to operators.

What’s next for H2O and Pani? “The long-term strategy is that AI is here to stay,” Bartlett says. He notes that H2O has vetted many software providers that claim to use AI and machine learning, only to discover that their products simply do linear regression, a statistical method that aims to predict the future based on the past. “We feel that Pani is definitely leading the industry in terms of AI and how that can help make water treatment plants smart.”

H2O envisions working with Pani to create plants that adapt to changing conditions—for example, by knowing what chemical dosage is needed to treat an influx of poor-quality water—and use energy more efficiently. “We’re not a software development company, so it’s hard for us to make those quantum leaps,” Bartlett says. “Pani is the shortcut,” he adds, “a huge stepping-stone to get that end result.”



Source link

You may also like