Building Smarter Water Systems

In the era of self-driving cars, digital assistants, and other smart things, can the same level of autonomy and “intelligence” be embedded in water systems? Such technologies have the potential to dramatically reshape adaptation to some of the greatest water challenges, such as floods and droughts. Software-updatable water systems are well within reach, promising to enable highly cost-effective water infrastructure that dynamically redesigns itself in response to changing needs and uncertain inputs.

Background

Recent news coverage has concerned droughts in the American West, contaminated tap water in the Midwestern United States, and floods in California, New England, and the southeastern coastal states. In fact, flooding is the leading cause of extreme weather fatalities in the United States (Vörösmarty et al. 2010).

At the same time, dry regions struggle to find new and clean sources of water. By many estimates, the capture of stormwater in Los Angeles could offset 10 billion gallons per storm (Monte 2015), a large portion of the city’s annual water budget. Unfortunately, most of this water is washed into the Pacific Ocean within hours after a storm. Capturing it requires distributed storage and treatment, both of which are limited.

The scarcity of water in the West and other parts of the United States is not helped by the age of the infrastructure that conveys it. By many estimates, nearly 20 percent of treated water is lost through old and leaky pipes, some of which were constructed at the turn of the last century (US EPA 2013).

Aside from water losses, leaks present significant revenue and energy challenges—treatment and conveyance across water systems account for almost 2 percent of the US energy budget (US EPA 2017). Thus, the need to better track and address challenges to water supplies and water distribution is imperative to maintain safe, leak-free, and energy-efficient water systems.

Status of US Stormwater Systems

Recent flash floods (e.g., Hunter 2016) are an all too common, dramatic example that aging infrastructure is struggling to keep up with changing and increasingly severe weather patterns. In addition, nutrients, metals, and many other pollutants are washed from urban surfaces when it rains, eventually ending up in streams, lakes, and oceans (Barco et al. 2008; Finkebine et al. 2000; Wang et al. 2001). And many parts of the country are dealing with chronically impaired coastlines due to algal blooms, which are driven, in part, by urban stormwater runoff (Carey et al. 2014; Doughton 2015; Wines 2014).

Infrastructure Challenges

To address flooding and water quality challenges, cities build and maintain complex networks of distributed stormwater assets such as pipes, canals, and basins. This infrastructure reduces flooding by moving water away from roads and buildings during storms. It also includes natural elements, such as wetlands and raingardens, to capture sediments and dissolved pollutants before they can be discharged to downstream ecosystems.

In some communities, however, stormwater and wastewater are combined, sharing the same pipes. With these systems, large storms can lead to sewer overflows into natural waterways, introducing viruses, bacteria, nutrients, pharmaceuticals, and other pollutants.

Many US stormwater systems were designed at times of less stringent regulations and for populations different from those they now serve. Most are approaching, or already exceed, their design life and face problems similar to those of America’s other ailing infrastructure. The American Society of Civil Engineers (ASCE) report card has given US stormwater infrastructure a near failing grade (ASCE 2017), and the problem is echoed in the National Academy of Engineering’s Grand Challenge to “restore and improve urban infrastructure” (NAE 2008).

Costly and Piecemeal Fixes

The distributed nature and massive size of most municipal stormwater infrastructure makes it impossible to dig up and resize the entire system. Instead, problems are often fixed one by one through expensive construction projects that are difficult to change afterward. As such, most stormwater systems comprise an amalgam of distributed fixes that have been constructed over decades and rarely add up to an optimized whole.

One of the country’s largest stormwater and sewer tunnels was recently built in Chicago, and a few -other cities are following this billion-dollar trend (Evans 2015). But these impressive, large construction projects are a luxury for most communities, many of which face challenges in simply maintaining their existing systems.

Even in small communities, stormwater systems can quickly add up to hundreds of linear miles of assets that must be maintained and repaired. In many US cities low or no fees are charged for stormwater services, in comparison to drinking water or sanitary sewer systems. With highly limited revenue streams, solutions that rely on new construction cannot keep pace with evolving community needs and uncertain weather. As cash-strapped communities seek more resilient infrastructure solutions, novel alternatives must be explored.

Technologies for Smarter Water Systems

The role of information technology in managing water supplies and drinking water distribution systems has been made evident in a number of applications. These include sensors for the management of large hydro-power and agricultural basins (Kerkez et al. 2012; Rheinheimer et al. 2016), real-time pump optimization and leak localization systems (Stoianov et al. 2007; Whittle et al. 2010), drinking water contamination detection (Ostfeld and Salomons 2004; Storey et al. 2011), and even residential WiFi irrigation widgets.

A burgeoning smart water industry is beginning to fill the needs of modern water utilities and municipalities. The overall technology outlook for water supplies and drinking water systems is promising, notwithstanding much work that must still be done in the development of water quality sensors for important contaminants, such as lead, bacteria, and viruses.

Not all water sectors are embracing technology at the same pace. This is particularly true across storm-water systems, an often overlooked and possibly the most poorly funded subset of urban water infrastructure (Kea et al. 2016). Instead of relying on costly construction, new technologies may make it possible to use existing systems much more effectively.

Industrial and academic efforts are under way to demonstrate the benefits of real-time sensing, computation, and wireless connectivity for the management of urban watersheds. The driving hypothesis behind this work is that smart stormwater systems will vastly shrink the size of infrastructure required to manage runoff pollution and other impacts of changing weather. This approach will dynamically repurpose existing stormwater systems by adapting them on a storm-by-storm basis.

Figure 1 

Researchers at the University of Michigan have been spearheading the development of open source technologies that enable watersheds to be retrofitted for sensing and real-time control (figure 1). These technologies and associated case studies are being shared through Open-Storm.org, an open source consortium of academic, industry, and municipal partners that seeks to provide a complete, “batteries included” template for the development of smart watersheds to combat flooding and improve urban water quality. The technologies include rapidly deployable wireless sensor nodes for the measurement of urban water flows, water quality, soils, and weather. These are complemented by cloud-hosted data services that allow measurements to be analyzed immediately, providing real-time information on the “health” of both the watershed and the infrastructure.

Commands can also be transmitted from the cloud to the watershed to change the configuration of infrastructure in real time. This is enabled by wirelessly controlled valves, gates, and pumps that can be quickly and easily attached to existing stormwater infrastructure, such as pipes, retention basins, or wetlands. Water levels at retro-fitted sites can be safely controlled to release water based on sensor measurements or real-time weather forecasts. By dynamically controlling flows across sites, system-level storage can be adapted to the unique nature of any given storm. This allows infrastructure to be “redesigned” or updated in near real time without the need for new construction.

Even a single remotely controlled valve can provide major benefits. At a basin or pond, a valve can be used to control flooding and reduce stress on downstream systems—a simple control algorithm closes the valve during a storm and opens it before the next storm starts. Storing water locally during a storm -temporarily removes it from downstream areas that may otherwise be prone to flooding. The captured water can be directly used for irrigation in adjacent neighborhoods or injected into underlying aquifers to replenish the ground-water table. Some level of treatment can even be achieved at the controlled site by promoting the capture of -sediment-bound or dissolved pollutants through settling or natural treatment. Many of these benefits can already be explored by cities and stormwater utilities through commercial real-time control solutions.

Perhaps the biggest benefit of real-time control is the ability to coordinate flows across entire infrastructure systems. For example, the upgraded large combined -sewer system in South Bend, Indiana, features over 100 sensors and 10 control points working in tandem to reduce sewer overflows over an area of some 40 square miles.

System-level control promises effective coordination of the many distributed parts of urban stormwater infrastructure at the scale of entire watersheds.

Testbed Implementation

Smart stormwater control networks are being deployed across the Midwestern United States as testbeds for system-level control. They were developed and deployed through the support of the Great Lakes Protection Fund, in partnership with the cities of Ann Arbor and Toledo, Washtenaw County, the Universities of Toledo and Michigan, and Michigan Aerospace.

Figure 2

The largest Open Storm testbed is in Ann Arbor (figure 2). The sensor network covers a 10 square mile urban watershed, in which some portions have a density of over 15 sensors per square mile. Multiple basins and wetlands have been retrofitted for control, sometimes in just one day by a group of university and high school students. Now in its second year of operation, the relatively inexpensive testbed has demonstrated the following:

• Measurements of soil moisture, water flows, rain, and water quality are transmitted in real time, analyzed, and made available to researchers and city engineers. This provides continuous performance insights that can be used to maintain or upgrade the existing infrastructure.
• Some of the largest controlled basins can store nearly 5 million gallons per storm, making it possible to control the majority of flows across the watershed.
• Real-time coordination of multiple valves has reduced flooding risk. Controlling how long water is held in basins after a storm has also reduced the output of sediments and nutrients from the watershed.

All of these benefits were achieved without new construction, but rather by using existing infrastructure more effectively.

Many scientific questions must now be addressed, but the testbed has proven to be effective and can serve as a blueprint for future smart watersheds. Researchers are also engaging with residents, city managers, and regu-lators on the value of these technologies to the community.

Simulating for Safety

Public safety demands that control algorithms be exhaustively validated and verified before control of watersheds is delegated to an autopilot. Efforts are focusing on achieving this through new simulation frameworks that allow large control networks to be modeled using state-of-the-art hydraulic solvers and control algorithms (Mullapudi et al. 2017).

Rather than running continuously, as is done in most water simulations, the water model can be halted after every step so that an external algorithm can “make decisions” on how to control valves or other assets. This allows a variety of algorithms to be evaluated in a physically realistic manner before being deployed on real-world systems (figure 3). These efforts are being freely shared through open source toolboxes to reach engineers in other disciplinary communities (e.g., control theory, machine learning) and encourage them to apply their own algorithms to combat flooding and improve water quality.

Figure 3

This simulation framework has already been used to begin investigating algorithms ranging across dynamical feedback control, market-based optimization, and reinforcement learning controllers. It can also help municipal managers decide on investments in real-time control.

Initial results obtained by University of Michigan researchers show that the addition of even a few control valves across urban watersheds allows the infrastructure to handle storms more than twice as large as those it was designed for. These findings also suggest that it may be possible to construct some new infrastructure at half the size when using real-time control, promising significant cost savings when compared to traditional methods.

Outlook

The adoption of smart water systems is no longer limited by technology (Kerkez et al. 2016), which has matured to the point that it can be ubiquitously deployed. -Rather, other barriers are becoming apparent and must be overcome to encourage and enable broader adoption:

• A new generation of control algorithms needs to be engineered, to ensure safe and autonomous operation at the scale of entire cities.
• More important, perhaps, social barriers to adoption need to be addressed. Doing so will require an understanding of how residents and decision makers perceive the benefits of smart versus traditional water systems.
• Finally, a new cross-disciplinary workforce needs to be trained and educated to be able to build, operate, and maintain this new generation of water systems.

New Control Algorithms

From a fundamental engineering perspective, there is a need to investigate how large water systems can be safely and autonomously controlled across the scales of entire cities and regions, requiring hundreds or thousands of control sites. To that end, extensive knowledge of water systems must be embedded in robust control algorithms.

While many control techniques for distributed systems have been successfully developed and applied in other engineering domains, an application-agnostic, one-size-fits-all approach may not be appropriate for water—what works for one type of water system may not work for another. For example, guidance on controlling flows across large spatial scales may come from research on reservoir operations (Wardlaw and Sharif 1999), water distribution systems (Cembrano et al. 2000), and control sewer systems (Marinaki and Papageorgiou 2005). However, the application of the same algorithms to stormwater may quickly reach limitations because of the need to account for challenges such as noisy sensor data, weather uncertainty, or the types of timescales, complexities, and feedbacks inherent in urban watersheds. Research is needed to determine which real-time control techniques will meet performance goals without risking the safety of nearby residents, property, and downstream ecosystems.

The need to solve these challenges presents an exciting opportunity for civil and environmental engineers to collaborate with colleagues in systems and computer science, electrical engineering, and other fields.

Public Buy-in

Even perfectly engineered smart systems may not be adopted if public trust is not secured. As autonomy moves into the sphere of residential and commercial water systems, it must be accompanied by a deep appreciation of how residents, decision makers, and -regulators perceive its benefits and risks.

The vast majority of water utilities and districts in the United States are publicly owned. An understandable level of risk aversion has developed over the decades since even small changes in operations can have large implications. The close connection between water and public safety, health, and other local priorities means that decisions are not always based on economics or efficiency. Engineers’ limited understanding of the perceptions of decision makers and the public regarding smart water technologies may be the biggest barrier to adoption.

The engineering disciplines must begin engaging with the social and economic sciences to develop a more holistic appreciation of the opportunities afforded by modern technologies for water management.

Workforce Development

Growing interest in smart cities promises exciting opportunities for a new workforce of engineers and technicians who will respond to a multitude of cross-disciplinary challenges. In the context of water, this will require educating a new generation of students whose knowledge of water systems will be combined with mastery of other disciplines, such as computer science and electrical engineering. Novel graduate and under-graduate educational initiatives are forming to fill this need, such as the University of Michigan’s Intelligent Systems graduate program in the Department of Civil and Environmental Engineering.

This new generation of students must be embraced by an industry willing to value skills that have not traditionally been part of its core. The economic and efficiency gains resulting from smart water systems will need to be translated into commensurate salaries for appropriately educated, technologically savvy engineers, who may otherwise be lured into much higher-paying tech careers.

Similarly, there is an opportunity to begin training a new workforce of smart water technicians and maintenance experts, who will not need college degrees to benefit from a tech career. This new workforce of engineering technicians may also help to advance equity and inclusion (Kuehn 2017).

Conclusions

Smart and autonomous water systems are well within reach, driven by pioneering efforts in a growing -number of US communities and utilities. Early implementation of these systems indicates that, before resorting to -costly new construction, it may already be possible to use existing infrastructure much more effectively.

Lessons learned by early adopters are being shared through new cross-sector groups and consortiums of regulators, companies, utilities, and academics, such as the international Smart Water Networks Forum (SWAN), and initiatives of established water organizations, such as ASCE’s Environmental Water Resources Institute (EWRI) and the Water Environment Foundation (WEF). In addition, the National Science Foundation, through its new Smart and Connected Communities program, is funding efforts to encourage fundamental discovery and meaningful engagement with cities and communities; one recently funded project will work across a number of US cities to investigate and overcome major barriers to adoption of smart stormwater systems (NSF 2017).

Anyone interested in efforts to enhance water management is encouraged to engage with these groups or to learn and contribute through open source efforts such as WaterAnalytics.org and Open-Storm.org.

References

ASCE [American Society of Civil Engineers]. 2017. -American Infrastructure Report Card. Available at www.-infrastructurereportcard.org/.

Barco J, Hogue TS, Curto V, Rademacher L. 2008. Linking hydrology and stream geochemistry in urban fringe watersheds. Journal of Hydrology 360(1–4):31–47.

Bartos M, Wong B, Kerkez B. 2018. Open Storm: A complete framework for sensing and control of urban watersheds. Environmental Science: Water Research and Technology. doi:10.1039/C7EW00374A.

Carey RO, Wollheim WM, Mulukutla GK, Mineau MM. 2014. Characterizing storm-event nitrate fluxes in a fifth order suburbanizing watershed using in situ sensors. Environmental Science and Technology 48(14):7756–7765.

Cembrano G, Wells G, Quevedo J, Pérez R, Argelaguet R. 2000. Optimal control of a water distribution network in a supervisory control system. Control Engineering Practice 8(10):1177–1188.

Doughton S. 2015. Toxic runoff kills adult Coho salmon, study finds. Seattle Times, October 8.

Evans MA. 2015. The sewage crisis in America: Flushing the toilet has never been riskier. The Atlantic, September 17.

Finkebine JK, Atwater JW, Mavnic DS. 2000. Stream health after urbanization. Journal of the American Water -Resources Association 36(5):1149–1160.

Hunter C. 2016. 6 dead in Texas floods, and more rain is coming. New York Times, May 30.

Kea K, Dymond R, Campbell W. 2016. An analysis of patterns and trends in United States stormwater utility systems. Journal of the American Water Resources Association 52(6):1433–1449.

Kerkez B, Glaser SD, Bales RC, Meadows MW. 2012. Design and performance of a wireless sensor network for -catchment-scale snow and soil moisture measurements. Water Resources Research 48(9).

Kerkez B, Gruden C, Lewis M, Montestruque L, Quigley M, Wong B, Bedig A, Kertesz R, Braun T, Cadwalader O, and 2 others. 2016. Smarter stormwater systems. Environmental Science and Technology 50(14):7267–7273.

Kuehn D. 2017. Analyzing the engineering technician and technologist workforce: Data coverage and gaps. The Bridge 47(2):11–18.

Marinaki M, Papageorgiou M. 2005. Optimal Real-Time Control of Sewer Networks. London: Springer Science & Business Media.

Monte M. 2015. DWP to unveil plan to capture storm runoff. LA Times, June 25.

Mullapudi A, Wong BP, Kerkez B. 2017. Emerging investigators series: Building a theory for smart stormwater systems. Environmental Science: Water Research and Technology 3(1):66–77.

NAE [National Academy of Engineering]. 2008. NAE Grand Challenges for Engineering. Washington: National Academy Press.

NSF [National Science Foundation]. 2017. NSF -announces $19.5M in awards to support fundamental research to advance the nation’s local cities and communities. -Alexandria VA.

Ostfeld A, Salomons E. 2004. Optimal layout of early warning detection stations for water distribution systems security. Journal of Water Resources Planning and Management 130(5):377–385.

Rheinheimer DE, Bales RC, Oroza CA, Lund JR, Viers JH. 2016. Valuing year-to-go hydrologic forecast improvements for a peaking hydropower system in the Sierra Nevada. Water Resources Research 52(5):3815–3828.

Stoianov I, Nachman L, Madden S, Tokmouline T, Csail M. 2007. PIPENET: A wireless sensor network for pipeline monitoring. IEE 6th International Symposium on Information Processing in Sensor Networks, April 25–27, -Cambridge MA, pp. 264–273.

Storey MV, van der Gaag B, Burns BP. 2011. Advances in on-line drinking water quality monitoring and early warning systems. Water Research 45(2):741–747.

US EPA [Environmental Protection Agency]. 2013. Water Audits and Water Loss Control for Public Water Systems. Washington.

US EPA. 2017. Energy Efficiency for Water Utilities. -Washington. Available at https://www.epa.gov/sustainable-water-infrastructure/energy- efficiency-water-utilities.

Vörösmarty CJ, McIntyre PB, Gessner MO, Dudgeon D, Prusevich A, Green P, Glidden S, Bunn SE, Sullivan CA, Reidy Liermann C, Davies PM. 2010. Global threats to human water security and river biodiversity. Nature 467(7315):555–561.

Wang L, Lyons J, Kanehl P, Bannerman R. 2001. Impacts of urbanization on stream habitat and fish across multiple spatial scales. Environmental Management 28(2):255–266.

Wardlaw R, Sharif M. 1999. Evaluation of genetic algorithms for optimal reservoir system operation. Journal of Water Resources Planning and Management 125(1):25–33.

Whittle AJ, Girod L, Preis A, Allen M, Lim HB, Iqbal M, Srirangarajan S, Fu C, Wong KJ, Goldsmith D. 2010. WaterWiSe@SG: A testbed for continuous monitoring of the water distribution system in Singapore. In: Water Distribution Systems Analysis 2010, eds. Lansey KE, Choi CY, Ostfeld A, Pepper IL, pp. 1362–1378. Reston VA: ASCE. Available at http://cedb.asce.org/cgi/WWWdisplay.cgi?284708.

Wines M. 2014. Behind Toledo’s water crisis, a long-troubled Lake Erie. New York Times, August 5.