A recent National Science Foundation publication calls UW chemical engineering professor Larry Ricker "a model citizen with his mind in the gutter." How can that be? Ricker has recently completed installation of an optimal control system for the city of Seattle's sewer network, and his computer models are already saving taxpayers plenty and helping the environment at the same time.
Like many other U.S. cities, Seattle has "combined" sewers that carry both domestic sewage and storm runoff. So when a storm pummels Seattle with heavy rains and the runoff exceeds the sewer capacity, raw sewage is released to the nearest body of water--Lake Washington, or Puget Sound, for instance. These releases are called "combined sewer overflows" or CSOs, and although progress has been made to reduce them, there is nevertheless a total of about 2 billion gallons of CSOs every year. Seattle is under the gun to eliminate them by the year 2005 to satisfy U.S. Environmental Protection Agency requirements.
With a grant from the National Science Foundation, Ricker and colleagues developed a control system that optimizes routing of sewer loads in order to prevent overflows and to maximize the proportion of sewage sent to the treatment plant.
At the beginning of a typical storm, sewer pipes are nearly empty, and there is considerable unused capacity in the system. As the pipes begin to fill, constraints of the system may lead to a CSO. Once the sewage depth in a pipe section exceeds a certain threshold, a CSO is inevitable. Moreover, limitations on the rate at which sewage can flow from one part of the network to another may cause a CSO at one location even though storage capacity is under-utilized elsewhere.
The new control system developed by Ricker and graduate student Marc Gelormino sends commands to automated sluice gates and variable-speed pumps, which regulate the flow at 23 points in the backbone of the network. These control points permit the accumulation of sewage in each upstream zone to be adjusted. As rain continues to fall, sensors provide instantaneous values of rainfall intensity, and a computer model converts these to predictions of flow rate entering each sewer zone. Those data, along with actual liquid-level and flow measurements in the pipes, allow the performance of the controller to be fine-tuned.
The Seattle sewer authority, Metro, installed Ricker's software for a trial run during the 1993-94 rainy season. It achieved a 10% reduction in CSOs, which Ricker believes can be improved still further. To achieve an equivalent reduction by constructing larger sewers or other facilities would have cost local taxpayers some $10 million. Ricker says the "model-predictive control" software remains in operation. It's an approach, he notes, that could help manage other networks, such as subways, dams, and irrigation systems.