Success on a Big Scale

A medium-pressure UV disinfection system provides cost-effective pathogen removal for a new 315 mgd facility in San Francisco.
Success on a Big Scale
A look inside the reactor shows the UV intensity sensors, sleeves, and lamps.

Interested in Disinfection?

Get Disinfection articles, news and videos right in your inbox! Sign up now.

Disinfection + Get Alerts

To comply with the U.S. EPA mandate to treat unfiltered water to 2.0 log Cryptosporidium removal, the San Francisco Public Utilities Commission built the $114 million, 315 mgd (design) Tesla Water Treatment Facility in Tracy, Calif.

Engineers tested low-pressure and medium-pressure UV systems at the original chlorination facility. “The pressure refers to mercury vapor pressure in the lamps,” says Enio Sebastiani, water quality engineer. “For a facility our size, medium-pressure 48-inch-diameter Sentinel Chevron UV reactors [Calgon Carbon Corp.] had the lowest overall construction and operational costs. They also treat higher flows with lower head losses.”

Steve Rotondo, Grades T5 and D5 certified journeyman operator, and a Calgon team led by Marc Frangipani, started the reactors in May 2011. “We had until April 2012, when the mandate became effective, to bring the units online,” he says. “During the process, Calgon made modifications that were later incorporated into its product line.” Tesla was the first installation of the new Chevron design.

From the mountains

Flows at the UV facility, the largest of its kind in the state, vary from 70 mgd in winter to 315 mgd in summer. It replaced a 300 mgd (design) chlorination facility treating unfiltered water for the Hetch Hetchy Regional Water System, serving 2.5 million Bay Area residents. Pristine water comes from Sierra Nevada snowmelt in Yosemite.

The facility includes sodium hypochlorite, carbon dioxide for pH correction and fluoridation before UV treatment using twelve 45 mgd stainless steel reactors set in two parallel trains of six. Each unit has four lamp banks totaling nine 20 kW lamps with Quickwipe wipers that mitigate quartz sleeve fouling.

Two power supply cabinets with programmable logic controllers automate all procedures. The computer divides the 45 mgd setpoint into the total flow rate, then activates the correct number of reactors, plus a backup. “At our maximum flow rate of 315 mgd, each train has two standby units,” says Rotondo. “We also set our kill target at 2.3 log and are achieving 2.4 log.”

Shakedown cruise

Rotondo, who spent 25 years in Air Force Civil Engineer construction, prepared for system startup by studying reactor overviews. “I was never exposed to UV treatment, so the learning curve was big,” he says. “I spent hours discussing strategies and issues with Frangipani and his team.” The first topic they brainstormed was the wiper assemblies, which didn’t extend and retract fully. An all-threaded rod moved the one-piece stainless steel plate back and forth, while the wire brush clamped to it cleaned the sleeves. They activated every 120 minutes.

“The units were factory tested, but not under actual flows,” says Rotondo. “Real-life applications revealed some concerns. Once Marc replaced the magnetic couplers on the rods with super-duty couplers, the assemblies worked well.”

The next gremlin was trickier to diagnose. Each lamp is assigned a well with UV intensity sensor, which reads the lamp’s radiance and sends it to the computer for comparison with the calculated percentage. Major variances indicate dirty sensors, wells or sleeves. Operators were doing so much cleaning that Rotondo kept a spreadsheet for three months, then emailed the documentation to Frangipani.

Frangipani’s team discovered condensation on the sensors as they entered the wells, and mostly on the devices’ etched serial numbers. His team created pockets in the 108 sensors, then they inserted desiccant packages to absorb moisture and maintain a clear line of sight to the lamps and sleeves. “The root cause of the problem was a unique combination of water temperature and humidity at the site,” he says.

Frangipani taught Rotondo how to calibrate the sensors, including how to adjust the unit’s scaling factor to compensate for distances between the grab sample and the analyzer. Then Rotondo trained the numerous co-workers rotating between the various East Bay Field Facilities. He used his military experience to maximize sessions. “Replacing three bulbs will take longer than replacing one, so I combined that exercise with teaching them how to drain and fill reactors,” he says. Bulbs can be replaced without draining the units.

Beyond standard design

Operators tested lamps to learn their limits. Most surpassed the 5,000-hour warranty by 2,000 hours. “We changed out some after 8,500 hours,” says Rotondo. “While initially testing lamp performance at higher hours, we were not comfortable running them any longer even though they gave no indications of failure.”

Tesla normally runs three banks per reactor, averaging 60 kW. Reactors running four banks use 75 kW and signal Rotondo that something is wrong. “Our raw water averages 89 percent UVT [ultraviolet transmittance] annually with turbidity between 0.2 to 0.3 NTU,” he says. “UVT determines how many lamps are on and how hot they burn. Hetch Hetchy water is so clean that the fourth bank runs only if something needs cleaning, has failed or is at lower UVT.”

Testing the UV system and construction upstream have pushed the equipment far beyond normal operational limits. One test ran all four banks in every reactor with only 10 mgd going through them. For four months, Tesla ran eight reactors in auto mode, and three backup units with the flow control valves manually locked open to protect pipeline workers up country from possible harm in case programming should close the valves.

“The system has responded well to every scenario we gave it,” he says. “It isn’t difficult to operate, it’s reliable, and as far as off-spec water, we’re hitting home runs every month.” To date, Tesla has not come close to the maximum allowable pathogen limit.


Comments on this site are submitted by users and are not endorsed by nor do they reflect the views or opinions of COLE Publishing, Inc. Comments are moderated before being posted.