South Bend, Ind., to Try Sensors to Halt Overflows

Engineers from Purdue and Notre Dame universities are working with Indiana startup EmNet LLC on a wireless sensor network for the city of South Bend to prevent raw sewage from overflowing into waterways, especially from surging runoff during storms.

The system will use a citywide network of 105 manhole-mounted sensors and "smart valves" to automatically hold back the flow of rainwater and sewage in existing sewer system pipes and retention basins until the storm has passed.

"To our knowledge, this monitoring system will be the first of its kind in the world because it will be the largest wireless sensor network in a permanent, industrial setting," said Luis Montestruque, chief executive officer of EmNet, founded in 2004 and located in Granger, Ind.

The system, called CSOnet, consists of numerous computer chips that communicate with each other over a wireless radio network. These microcomputers are embedded in the city sewer system and are connected to flow sensors, pressure sensors, and valves in a network that works in a cooperative manner to control storm runoff, Montestruque said.

Sensors are mounted on the undersides of manhole covers and will have to perform in an urban setting full of interference sources, said William Chappell, a Purdue assistant professor of electrical and computer engineering who helped design the sensor technology.

"The sensors must be made to operate in harsh conditions and adapt dynamically to changes in the wireless system, such as interference or the presence of parked or moving cars," Chappell said. "And the system will need to broadcast sensing data generated underground to a network that operates above ground in a challenging environment."

The system is expected to be fully functional next summer, with work progressing in stages as the system evolves to increase its monitoring capabilities, said Gary Gilot, South Bend's public works director.

Research to develop the system began in 2004 under the direction of Jeffrey W. Talley, an associate professor of civil and environmental engineering at Notre Dame. Talley led a team of project participants at Notre Dame, Purdue, the city of South Bend, and EmNet that in 2005 deployed a small prototype of the CSOnet system in the city. EmNet later took over the project to expand the earlier prototype into a citywide system.

The technology is an example of a "cyberphysical system," or a network of computers tightly integrated with sensors and motorized controls, said Michael Lemmon, a Notre Dame professor of electrical engineering.

"There are many sensor networks in operation around the world, but few attempt to do active control, and that's one of the innovative aspects of this project," said Lemmon, who is leading work to develop and test computer algorithms that enable CSOnet to control storm runoff in an optimal manner.

Saurabh Bagchi, a Purdue assistant professor of electrical and computer engineering, has developed critical software, called middleware, for the system. The software allows the sensors to talk to each other in an "ad-hoc mesh network."

"The ad-hoc part means there is no need for pre-existing infrastructure," Montestruque said. "It doesn't rely on a cell phone tower or telephone lines for the wireless portion of the communication to work. The mesh part means that between every two devices there are a number of different paths for it to communicate, bypassing interference and obstacles."

Unlike other wireless systems, the network does not require a command center and can be reprogrammed wirelessly from a remote location. Information from the sensors will be relayed to a server operated by EmNet.

The sensor data will be used to monitor hydraulic conditions in the sewer system, indicating when excess runoff and raw sewage are about to overflow. Then valves will divert the flow into temporary storage sites.

The sewage will be selectively released later so that it flows into the treatment plant when capacity is available, preventing the waste from being dumped into Indiana waterways, including the St. Joseph and Wabash rivers.

The system being developed for South Bend will control flows at about 20 locations in the city sewer system. Algorithms developed by Lemmon allow each of these locations to make flow-control decisions in a manner that requires only information from neighboring points, an approach that ensures optimal control of stormwater flows while greatly reducing the amount of information that must be exchanged between different locations in the system.