Chapel Hill and Carrboro can use nine million gallons of water a day. Pumping that much water is a big job—and an expensive one. With some savvy calculations, graduate student Amy Buege figured out how to trim Orange County’s electric bills and keep clean water flowing.

Every day the Orange Water and Sewer Authority (OWASA) sends water where it’s needed, when it’s needed, pumping six to nine million gallons from University Lake and Cane Creek into sinks, bathtubs, and washing machines in Chapel Hill and Carrboro.

But pumping water can be very expensive. Electricity costs alone can run hundreds of dollars each day. Staff at OWASA realized that using electricity more efficiently would help hold down operating costs. The next step was to figure out how.

OWASA turned to the Department of Operations Research at UNC-CH, which specializes in studying the most efficient use of resources in large-scale operations. Through its hands-on course in operations research practice, the department matches graduate students with Triangle-area organizations that request help solving problems. For OWASA, the department assigned master’s student Amy Buege.

Buege realized that pumping water at a constant rate throughout the day is not very cost-efficient. During the peak hours of use each day, electricity costs three times as much as it does during hours of lower usage.

“Right now, they pump pretty much according to need,” Buege explains. “And the hours of greatest need are often the hours when electricity is most expensive—between six a.m. and one p.m. in the winter and between one p.m. and nine p.m. in the summer.”

To Buege, is seemed clear that the best pumping strategy would be to pump as little as possible during these peak-use hours, yet ensure that the community received all of the water it needed each day. But while the strategy was simple, calculating the most efficient pumping schedule was not.

Buege attacked the problem using dynamic programming, a technique that takes a complex problem that changes over time and breaks it up into simpler stages. Starting with the final stage, dynamic programming works backward through intermediate stages toward the beginning, optimizing cost and efficiency along the way. Each stage requires detailed computations.

“You know the end result,” Buege says, “so then you work backwards to find the cheapest way to get to that end.”

Buege knew the tanks must be full by 6 a.m., when electricity can be most expensive and when people get up to shower and shove off to school or work. She also had to keep in mind that the tanks must have a certain amount of water in them at all times, a minimum that the plant operators call the “comfort level.” “We want to err on the side of surplus,” she says.

Buege mapped out a daily pumping strategy that minimizes energy usage during the most expensive hours. Her optimal solution calls for OWASA to pump to the comfort level—maintaining water-quality standards and sufficient water supply—during hours of high electrical cost. During the rest of the day, the plant can pump to capacity.

But a daily schedule must be flexible, Buege says. If students leave for spring break or if a heat wave prompts people to fill their pools and water their lawns, OWASA must respond to the changing need. So Buege’s solution requires plant operators to adjust their pumping schedule according to the expected daily demand.

Water emergencies such as line breaks or fires are also part of the plant’s daily concerns. And they can be expensive. In addition to hourly electrical costs, the plant pays a monthly demand charge, a fee based on its maximum electrical usage during the month. One emergency can drive up electrical costs significantly by increasing the demand charge for the entire month.

So Buege recommended that the plant use its own generators to produce the extra energy it needs in an emergency. If OWASA uses on-site generators to pump to capacity when a water pipe breaks, its monthly demand charge won’t increase.

Buege’s project resulted in a manual for plant operators, outlining daily pumping schedules that vary according to the expected daily demand. If the manual is followed, the plant could reduce its electrical use during the most expensive times of the day by as much as 60 percent. It’s a money-saving tactic that could be easily implemented by plant operators. “Until now,” Buege says, “no one at the plant had ever really looked at the cost of pumping in terms of the energy that’s required.”

OWASA plans to follow many of the guidelines Buege has recommended, says William “Ed” Kerwin, Jr., executive director of OWASA. They may use others after the plant gets new equipment, such as an additional water storage tank. “Amy did a fantastic job,” Kerwin says. “Her technical work as well as her interactions with our staff were excellent, all of which should allow us to work a little smarter and save our customers some money.”

What is Operations Research?

For most of us, the term “operations research” requires some explaining. Operations research is the science of optimal decision making, applied primarily to the complex activities of businesses, governments, or other organizations. Using mathematical models and computers, operations researchers gather information about an organization’s operating environment, calculate the efficiencies of various management options, and provide guidelines for people making decisions about scheduling, routing, production, inventory management, resource distribution, and many other kinds of operational issues. Operations research has increased the efficiency of large industries, including those in telecommunications, manufacturing, and transportation. But it has also proved useful outside of industry—in farming, in health care, and in the management of fisheries and water resources.

Because operations research is applied to actual management problems, studies at UNC-CH have become directly involved with public service. Each year, Carolina’s Department of Operations Research—a small group of six faculty members and about 30 graduate students—works with government agencies, businesses, or units within the university, helping to improve their procedures. During the spring semester, graduate students work with clients who have submitted problems for consideration. For example, students have offered plans for relieving congestion at Glaxo-Wellcome, handling incoming calls at the Chapel Hill police department, and scheduling in the Durham Police Department.

On campus, operations-research projects have helped answer a number of questions involving large investments of time and money. One study provided recommendations about the most cost-effective schedule for replacing faculty members’ computers. Other projects have helped guide the university’s recycling program. According to Herbert Paul, director of the physical plant, work by graduate student Karen Susenna laid the groundwork for the university’s recycling program as early as 1988. By selling its recyclable materials and reducing tipping fees at landfills, the university now offsets much of the cost of waste collection and disposal.

“Karen’s efforts more or less gave us the framework we use to move our material to market,” Paul says. “She did an excellent job.”