With water age, operators are concerned with disinfection by-products formed when organic residue that lingers after treatment reacts with chlorine, creating a gradual accumulation over time. The fix is to not let water stay anywhere in the system for more than a few weeks, at most. The way to get a handle on it is to model the age of the water in each tank and pipe, track it through the system and then strategically open valves and hydrants to flush the old water out, if necessary.

"Through the day, we can run 'predicted' compared to 'output.' That can signal problems in the system, like a closed valve," Jacobsen says. "We have had valves slam shut and have had partially closed transmission lines. This gives us model calibration every day, so our confidence in the model increases over time as we work with the system."

Having high-speed computing and confidence in the model opens the door to analysis tools gaining wider use. Some use genetic algorithms that approach complex problems by assuming a solution that should match the data and, then, re-solving the problem again and again with slight variations to find a result that best fits the data through an evolutionary process. Bentley's analysis tools built around this approach are named after Charles Darwin, but genetic algorithms are being used by all the major vendors now. They include software for calibrating the models themselves; for finding the probable locations of leaks; for optimizing pump use to ensure not only cost efficiency, but service and water quality; and recommending pipe replacement plans that factor in not only pipe age and materials but also service history.

Some of these, such as pump optimization, can be used on simple systems with scanty data and still yield valuable results, Jacobsen says, which makes them potentially important tools for optimizing systems in the developing world, where water may be precious.

"It is amazing the amount of insight you can get from a mathematical model once you get it fine-tuned," says Jacobsen.