Math Magic

Technology lets water agencies use data and analytics to improve system pressure management and detect losses due to leakage
Math Magic
Using IBM advanced analytics, water loss is reduced by optimizing the setting of pressure-reducing valves based on data from existing sensors as well as other sources.

Interested in Instrumentation?

Get Instrumentation articles, news and videos right in your inbox! Sign up now.

Instrumentation + Get Alerts

Water utilities use a variety of technologies to measure and optimize pressure on their systems and detect leaks. Now IBM is testing a technology that lets utilities perform both functions, largely using data they already collect from in-the-field instruments.

The analytics-driven leak detection and pressure management technology is in pilot testing at Valley of the Moon Water District (VMWD), a purchaser of water from wholesaler Sonoma County (Calif.) Water Agency (SCWA). Another new pilot is in place with the City of San Diego (Calif.) Public Utilities, in partnership with the Richard Brady & Associates consulting engineering firm.

Developed by scientists at IBM Research, the pressure management solution provides recommendations for water pressure adjustments based on usage, weather, and environmental conditions. The benefits of improved pressure management include reduced water loss, energy savings, reduced wear on the infrastructure, and improvement in the quality and turnover of stored water.

Peter Williams, chief technology officer with IBM Big Green Innovations talked about the technology and its potential benefits and applications in an interview with Water System Operator.

wso: What exactly does this offering consist of?

Williams: There are a number of aspects to it. The idea is to enable water agencies to identify opportunities to optimize pressure, thereby potentially reducing energy (to the extent that their pressure is not gravity-generated), and also reducing wear and tear on the infrastructure.

Recent data shows that for every 10 percent reduction in maximum pressure, you reduce the likelihood of leaks by 14 percent. That obviously is worth doing. At the same time, the technology is designed to enable utilities to detect leaks, even where they have not broken the ground. It helps them understand where and how big the leaks are.

wso: Why would a water system be under excessive pressure?

Williams: Many water agencies somewhat over-pressurize their systems in order to guarantee acceptable pressure at the customer end. If there are opportunities to reduce pressure at certain places in the system while still maintaining acceptable pressure, then the tool we’re creating will identify that.

wso: In basic terms, how does this technology work?

Williams: We start with a hydraulic model, which if a water agency has one today, is probably a back-office tool that they deploy perhaps once or twice a year to run a few scenarios. What we do in effect is run that hydraulic model continuously, so that it provides a continuous operating baseline for what the system “should” be doing.

We then compare the model with data coming from instrumentation on the system — bump meters, flow and pressure transducers and so on. And we run a set of statistical algorithms and look for anomalies between what the hydraulic model says ought to be happening and what is actually happening, based on data from the field, usually coming through the agency’s SCADA system.

We then run other statistical algorithms to analyze those anomalies. For example, for leak detection, where we get a variance, we use an optimization tool to identify the location and the size of the leak that best explains the variation. The pressure detection is a different optimization calculation, but it’s the same basic idea. Given all the known constraints, what pressure does there have to be in segment X of the system to deliver acceptable pressure at the end user connections that feed off that segment?

wso: Where does the necessary instrumentation come from?

Williams: In theory, the water agency already has that equipment, although in both of our pilots we have provided additional instrumentation. The more measurement points you have, the more granular you can get with your detection capabilities. There’s not much point in knowing you’ve got a leak in a 5-mile stretch of pipe. You want to be able to narrow that down to a few hundred feet if you possibly can.

wso: What is the advantage of this technology over various other leak detection methods already in wide use?

Williams: Listening devices and other physical tools can be good solutions, although some surveys show that they fail to detect a large number of the leaks that are out there. Also, leak detection solutions don’t tend to allow the user to optimize pressure at the same time, and vice versa. What we’re providing is something that does both.

The technology can be integrated with a SCADA system or integrated through some other dashboard. If the water system is sufficiently instrumented, the agency can get the service and the value we provide with the same data they’re already collecting.

Furthermore, many water agencies are looking hard at advanced metering infrastructures (AMI), and AMI data is a wonderful source of additional information that will allow our solution to work even better.

wso: How does excessive pressure negatively affect infrastructure?

Williams: In two primary ways. If pressure is generated by pumping, then you’re generating more pressure than you need, and you’re wasting energy. And between 3 and 5 percent of the electricity consumption of the entire United States is used for moving water around. Water is heavy, and pumping is extremely energy-intensive.

The other impact is that if you’re over-pressurizing, you’re putting additional wear and tear and strain on the infrastructure — mainly at the joints. Pressure will find the weakest points on the system, which are usually the joints. If you can reduce the pressure, then all things being equal, and over time, you should be able to reduce the propensity of the system to develop further leaks.

wso: What was the origin of your pilot project with SCWA and VMWD?

Williams: We have had quite a long and productive relationship with SCWA, which has a two-tiered structure. SCWA has long been keen to improve collaboration with its retail contractors, of which VMWD is one.

They realize that if they are going to be collectively successful in managing the water resources in their area, which come mostly from the federally protected Russian River, they need to share information. They all need to understand what’s going on in the system and have one version of the truth. We’ve been working with them for a couple of years to provide tools that enable them to share information. The leak detection and pressure management work we’re now doing is an outgrowth of that.

wso: How would you characterize the pilot project results to date?

Williams: We started with pressure management. The VMWD chief engineer is on record saying he’s been doing his job for 30 years, and that we’ve come up with options for meeting customer demand he never would have thought of. The simple reason he never thought of them is that he’s not a computer — he’s a human being.

We’ve identified new ways for them to move water around the system. For example, they were having trouble getting a particular tank to fill with the available pressure. We helped them reconfigure the way they operate the valves and pumps on the system, and that problem has now gone away.

They are now able to maintain tank levels and the tank turns more efficiently than they expected. We haven’t quantified the energy savings that come from that yet, but it’s reasonable to expect that the savings will be there.

Right now, we’re doing leak detection in one subdivision within VMWD. The idea is that once we prove this system works, we will show that it works in the entire district, which has about 2,000 end customers. The next step will be to convince SCWD’s other retail contractors that this is a worthwhile thing to do.

wso: What has VMWD been doing previously for leak detection?

Williams: They had been mostly relying on public reports, and where they had leak-prone areas, they sent crews out specifically to hunt for leaks. That obviously is very labor-intensive.

wso: How would you assess the potential for this technology?

Williams: If you believe the U.S. EPA estimate, non-revenue water in the United States is 14 percent. The bulk of that is from leakage, and in some cities the percentage is very significantly higher than 14 percent. We see potential for this technology to help agencies get a handle on leaks and optimize pressure from inside the control room.

We know there are other related and analogous technologies available, and they all work differently. It will be interesting as we go down the track to figure out how to make some of those other technologies work with ours, and vice versa. We think this technology makes a major contribution to the management of water systems.



Discussion

Comments on this site are submitted by users and are not endorsed by nor do they reflect the views or opinions of COLE Publishing, Inc. Comments are moderated before being posted.