IBM Research: Smart water analytics

"It took a crew of technicians and special equipment to block off traffic, climb down under the road, and make the adjustments to the water pressure valves," said Segev Wasserkrug, lead researcher for IBM's First Of A Kind (FOAK) project that brought smart water analytics from Haifa, Israel to Northwest California's Sonoma County. "This was a real experience in 'physical meets digital'."

"Finding a better way for engineers to manage the pressure in a water network consisting of pressure valves, pipes, pumps, tanks, and sensors was no simple matter," explained Wasserkrug. "If there's too much pressure, more bursts are expected, and any small leak will result in even greater water losses. When the pressure is too low, tanks may not fill to the proper level and people may have problems using their taps and showers."

Before working with IBM, engineers from Valley of the Moon Water Distribution Agency (the organization that covers Sonoma County's water management) had to manually adjust the pressure of each valve and then wait for daily readings to see how each adjustment affected other areas of the water system. Getting all 10 valves adjusted to maintain optimal pressure across the system was a time consuming and complex process that was only done twice a year -- when the seasons change from winter to summer and back to winter.

"In the summer, we see increased demand for water because of the warm weather associated with filling swimming pools, summer water activities, and the need for more irrigation," said Wasserkrug. "In the transition to winter, we see a 30 to 40 percent reduction in water use because it rains, people don’t need to irrigate the land as much, and winter activities don't typically require much water."

Now, IBM analytics provide Valley of the Moon engineers with detailed information on optimal settings for each valve based on what’s happening across the entire system so they can be adjusted as necessary. And the IBM solution is unique in its ability to manage simultaneous changes for multiple valve settings.

"I think both sides gained considerable insight. For me, going out into the field and getting readings from actual sensors was very different from seeing data on the computer. The water engineers gained confidence in our system's analytics as we began to see the results from each subsequent set of adjustments," said Wasserkrug.

"We’re very pleased with the recommendations of the system. I've been operating this system for 30 years, and I never thought of making the changes the IBM system recommended," said Paul Gradolph, Operations and Maintenance Supervisor from Valley of the Moon Water District.

“Our ongoing collaboration with IBM is a clear indication of how the innovative use of technology helps us effectively manage the resources in our care.”

The new pressure management solution transformed the water network into a proactive system instead of one that was reactive. Beyond simply tracking the data, the analytics can identify trends in demand and help the engineers anticipate upcoming changes. This, in turn, means the engineers can make adjustments in advance and prevent problems such as pressure spikes before they occur.

The IBM Research - Haifa team is now working on a generic solution that can be applied anywhere, taking into account additional types of equipment such as pumps, as well as different variables such as pumps, frequency of pressure changes, and other unique water system characteristics. The next step with Sonoma County is to develop a solution that automatically identifies leaks based on mathematical models that compare the water entering the system and the amount coming out at various locations.

“Every change we make to the system must be designed to bring about real improvement and not something fleeting," said Wasserkrug. "We have to remember that we are influencing one of the basic human needs – the water we use and drink.“

Smart water analytics – when IT meets H2O