When the warning sirens sound in the chlorine chamber of the District's sewage treatment plant--a facility that holds one of the region's largest supplies of toxic chemicals--city records show the workers there often have a simple reaction: They disconnect the alarm.

It is one of many routines at the squat, brick structure known as Chlorine Building I that threaten not only the around-the-clock operations at the Blue Plains Sewage Treatment Plant in Southwest Washington, but also potentially the health of thousands of people who live or work in the area, said three former or current safety technicians assigned to the plant.

At least 180 tons of chlorine is stored at Blue Plains--some days there is as much as 630 tons--in a liquid form so toxic that even if only a portion of it were accidentally released, it could kill plant workers in seconds and create a poisonous plume more than 30 miles long.

Yet the plant's chlorine gas sensors are years past their normal replacement dates. Plant workers sometimes leave the chlorine building unattended, including once in June when a gas leak went undetected until a technician happened to walk in. On Wednesday, a cabinet for emergency breathing equipment next to two rail cars holding liquid chlorine was filled not with oxygen tanks and masks, but with rusty tools. And a panel of warning lights in Chlorine Building I is so unreliable that workers ignore the randomly blinking emergency lights.

In all, the technicians say, it is a disaster waiting to happen.

"You can only go on the grace of God but for so long," said James J. Bobreski, an engineering technician who last week was laid off by a firm that does consulting work at Blue Plains after he raised objections about the plant's safety.

"It is going to come back and haunt you. You are going to get caught."

The District has been warned repeatedly over the past six years by the U.S. Environmental Protection Agency about safety shortcomings at Blue Plains. Four years ago, the D.C. Water and Sewer Authority hired Bobreski's former firm, New Jersey-based Givoo Consultants Inc., to help monitor chemical safety at the plant.

But the problems persist, and late last month, a Givoo supervisor told Bobreski that he feared the firm might lose its bid for a $1 million-a-year contract renewal with the District if it kept badgering the city about safety flaws.

The supervisor did not know his remarks were being recorded. A copy of the recording, which was legal under D.C. law, was obtained by The Washington Post.

"There are certain things that need fixing, such as in the Chlorine Building," Givoo supervisor Dan G. Juanillo Sr. says on the tape. Water and Sewer Authority officials "know about it. . . . We have already written it up many times. We made a mention of it. Enough said. . . .

"What I am saying here is, we are stirring up a can of worms and it is going to [mess] up somebody somewhere and somebody is going to pay for it," Juanillo says on the tape. "And what we are looking at is Givoo right now."

Juanillo acknowledged in an interview this week that he made the remarks. Top Givoo officials at the company's Cherry Hill, N.J., headquarters said yesterday they were "very disappointed" and "disturbed" to hear what Juanillo had said, adding that Givoo's policy is to report all safety problems found by its technicians.

Top officials at the D.C. Water and Sewer Authority said Wednesday they had been unaware of such safety lapses at Blue Plains before being asked about them by The Post.

They began an investigation of chlorine safety at the plant last week, D.C. Water and Sewer Chief Engineer Michael S. Marcotte said. "I am very troubled," Marcotte said, noting his office is near Chlorine Building I. "If there are shortcomings, we will be working around the clock to get them fixed."

Marcotte said he believes that the primary cause of the problem was that Givoo employees were not doing their jobs and that Bobreski shoulders some of the blame. But he said the Water and Sewer Authority deserves part of the blame as well.

"If these happened on our watch, I certainly don't blame anyone else other than the folks who are currently working here, including myself," Marcotte said.

Marcotte and John Moore, a top official at Givoo, said they do not believe that the operations at Blue Plains threaten its employees or anyone else in the area.

"Are we on the brink of a safety crisis? Far from it," Moore said.

But D.C. Mayor Anthony A. Williams (D) said yesterday that he was "deeply concerned" by the questions of safety at Blue Plains and that his administration would immediately investigate the plant's chlorine facility.

"Citizens depend on government to monitor and prevent the release of potentially hazardous substances. It is an obligation I take very seriously, and I want to assure the public that this situation is being addressed immediately," Williams said.

"I have made it clear to all involved I want immediate action to provide the required safety. . . . I have directed Peter LaPorte, my director of emergency management, to begin a review of all safety issues at Blue Plains and recommend immediate action if problems are present. I expect a report from Mr. LaPorte within 48 hours."

Mikal Shabazz, director of the EPA's regional office of chemical accident prevention, said the agency also plans to investigate Blue Plains.

Useful and Dangerous

Chlorine--which is greenish-yellow when it is a gas and a clear amber when it is a liquid--is commonly used to purify drinking water, sanitize swimming pools and bleach clothes.

Its industrial uses are much broader; it is used to make everything from car upholstery to golf bags, wallpaper and window frames, as well as pharmaceuticals and medical equipment. For decades, sewage treatment plants across the nation also have used chlorine to help sanitize wastewater before it is discharged back into rivers and bays.

The chlorine used to wash clothing or disinfect pools is much less hazardous than the substance used at Blue Plains, which processes 370 million gallons of sewage a day for the District as well as parts of Fairfax, Loudoun, Prince George's and Montgomery counties.

The liquid chlorine used at Blue Plains, if released into the atmosphere, would quickly turn into a gas and expand as much as 460 times, forming a cloud that would hug the ground, because it is heavier than air. At just three parts per million, the resulting gas can cause eye and respiratory track irritation; it is fatal starting about 430 ppm. In the area immediately around a catastrophic spill--such as a chlorine rail car losing its entire load--the level can quickly climb to more than 20,000 ppm. The substance is so powerful that it was used as a chemical weapon by Germany during World War I.

The diluted household version of chlorine emits a penetrating odor if spilled, but simply evaporates instead of forming a toxic cloud. For this reason, many wastewater treatment plants have switched to a bleach-like version of the disinfectant.

The District has long known about the risks associated with its chlorine stockpile. A 1982 study commissioned by the city concluded that if one of the 90-ton chlorine tankers stored at the site were to rupture, people within 3.4 miles could be at risk of death, a ring that takes in Alexandria, Reagan National Airport and much of Anacostia. People first would notice an irritation to their eyes and throat perhaps as far as 31 miles away and the odor would be detectable even farther away, the study said.

The risk of a catastrophic rupture of a chlorine rail car is remote: About 50,000 rail cars of chlorine are shipped nationwide each year, but since 1960 only 10 people have died in rail car-related accidents, according to the Chlorine Institute, a trade association.

But smaller-scale leaks have occurred more frequently across the nation, including one at Blue Plains four years ago that injured four Potomac River fishermen who were overcome by a toxic cloud. At least two small leaks occurred this year at the plant, according to city documents and officials, neither of which has previously been disclosed.

An emergency plan prepared by the city this year concluded that a 10-minute leak in one of the hoses that feeds chlorine from the tankers to the wastewater treatment unit at Blue Plains could spread a toxic plume of chlorine one-third of a mile around the plant, an area that includes the Navy Research Laboratory. Nearly 4,000 people work in that area.

The commanding officers at the research laboratory and nearby Bolling Air Force Base have been trying for nearly a decade to persuade the city to replace liquid chlorine and other potentially dangerous chemicals at the plant with less hazardous options. The D.C. government has been studying that idea since at least 1992, but the effort has not gone much beyond the planning stage.

"The transport of chlorine and sulfur dioxide through our installations poses a significant safety and security threat to the military and civilian personnel who live and work here, as well as to the residents of the District of Columbia," said the 1991 letter sent to then-Mayor Sharon Pratt Kelly by the Navy and Army officials.

Marcotte, who has worked at the water authority for two years, said the agency recognizes the importance of the $18 million chemical conversion project, which is now being accelerated. But completion of the project is at least three years away.

Aging Sensors

Because of the dangers associated with liquid chlorine, the federal government has a long list of regulations on how it should be transported and handled and how those who deal with the chemical should protect themselves and the public.

The D.C. Water and Sewer Authority has told the U.S. government that such measures are in place at Blue Plains, and EPA officials said that until this week, they had no reason to believe otherwise. But city contractors, documents and conditions visible during a tour of the plant this week raise doubts about whether the local authority is living up to its promises.

Sensors installed to detect leaks, the alarm system designed to warn of potential problems, and the safety equipment--such as oxygen masks--that workers must rely on in emergencies all have been compromised, according to city records and the contractors. Meanwhile, internal memos at Blue Plains question whether Water and Sewer Authority staff members at Chlorine Building I pay enough attention to their jobs.

The first line of defense against an accidental chlorine leak are seven highly sensitive devices designed to detect even trace levels of chlorine. The idea is that if the plant's employees are quickly alerted to possible leaks, they can fix problems before they get out of hand. The sensors trip an alarm system that is supposed to have flashing lights and a loud siren.

The EPA warned the city in 1993 that during an inspection, auditors had found only one of four sensors checked to be in working order. Givoo Consultants has been paid about $4 million since 1995, according to city records, to make sure the sensors and other safety equipment work properly.

But city documents indicate that problems with the sensors have continued.

To begin with, the sensors need to be replaced about every two years, said George Kugler, vice president for research and development at Scott/Bacharach Gas Detection Products, the Pennsylvania-based manufacturer of Blue Plains's sensors. The two-year lifetime is outlined in the sensors' operating manual, but as of Wednesday the sensors at Chlorine Building I were, on average, more than four years old, with one dating to 1990, a Givoo technician said.

"That is not good," Kugler said. "We would not expect them to last that long."

The manual also recommends two tests to ensure the sensors are working properly: a detailed semiannual inspection using a sample of chlorine gas and a more basic, twice-a-month checkup. Neither test has been performed properly at Blue Plains in more than a year and perhaps longer, two Givoo technicians said this week.

The technicians said that they have been checking the sensors every other week but that the procedure they have used--which is not consistent with what the manual calls for--is much more likely to give a false indication that a inoperative sensor is working properly.

The technicians maintain they did the tests as instructed by water authority work orders; authority officials say it was Givoo's responsibility to do the tests as described by the sensors' manufacturer.

On Oct. 22, frustrated with the lack of response by the city and Givoo to his complaints about the sensors, a Givoo technician submitted a written report indicating that none of the seven sensors responded within the five-second standard established by the manufacturer. And four of them did not go off even after five minutes of chlorine exposure, his report said.

An order for four new sensors was placed last Wednesday--costing a total of $1,625--several days after The Post began its inquiry into the plant's troubles.

But the safety system's weaknesses go beyond the sensors.

A Chlorine Building control panel--filled with meters and lights that are supposed to signal possible trouble in the chemicals and in the wastewater as it passes through the plant--is considered unreliable by workers, said William Albrittain, manager of the plant's primary treatment.

"We basically ignore them," he said Wednesday morning as 12 lights on the nearby panel glowed, indicating high pressure, low pressure and other potential chlorine problems. Instead, he said, the workers inspect the equipment manually, though they do so just once an hour.

The building's alarm system also is unreliable. At least four times in recent months, technicians said they found electrical wiring for the audible alarm disconnected, which they blamed on plant workers who apparently were bothered by the noise from frequent alarms.

A Sept. 10 memorandum from a Givoo technician to his supervisor detailed the problem: "The chlorine building is one building that should have all of its alarms functioning at all times. . . . Failure to correct this problem could lead to death or serious injury, along with certain legal liabilities."

Even when the alarms and sensors are working properly, warnings aren't always noticed.

On June 30, a technician heard an alarm sounding as he arrived at Chlorine Building I at 7 a.m. No one was at the controls, and he discovered a gas leaking from a chlorine-mixing device. An emergency ventilation system--designed to suck toxic air out of the plant and neutralize any poisons--had automatically switched on. But it was ineffective, because the worker who had left the building earlier had propped open a door.

Other problems evident at Chlorine Building I this week or detailed in recent memos included:

* A supply cabinet next to the chlorine rail cars that is designed to hold emergency breathing equipment--which workers would use to try to contain a leak or leave the area in the event of a spill--held only two rusty pipe wrenches on Wednesday. A December 1993 letter sent by the EPA to Blue Plains Manager Walter F. Bailey warned the city about the need to have emergency oxygen tanks there.

* Bare wires and parts of the electrical system at Blue Plains are exposed in some places, leaving them vulnerable to short circuit.

* The roof at Chlorine Building I leaks, windows are broken, and pipes and valves that carry chlorine gas and other substances through the plant are rusted.

Each of the Givoo consultants interviewed said that in their years working at various plants--from plastic factories to other municipal wastewater operations to nuclear power plants--they have never seen conditions as hazardous as those at Blue Plains.

"I was frightened," said one technician who asked not to be identified, adding that he decided earlier this year to quit and find another job. "I said to myself, 'It is not worth working here for me. I don't feel safe here.' I wondered: Who is running this place, and how could they allow this to happen?"

Explanations Differ

D.C. officials and Givoo gave conflicting explanations this week about who is to blame for the faulty safety systems.

Juanillo, the Givoo supervisor whose conversation with Bobreski was taped, said in that conversation that the company had told water authority managers about the problems and noted that the firm recently proposed installing a new alarm system.

But Juanillo warned his colleague not to push the city too aggressively. Givoo is seeking an extension of its city contract, and on the tape Juanillo tells Bobreski he does not want to raise questions about why the city had not addressed safety problems at Blue Plains.

"I am not trying to say we are trying to cover it up," Juanillo tells Bobreski on the tape. "There has to be a different way of doing it without causing this kind of a problem."

Juanillo goes on to tell Bobreski that Givoo's responsibility is to tell the water authority about safety problems, and if they aren't corrected, it is the city's fault.

"We don't need to keep ramming it down their [expletive] throat," Juanillo says on the tape.

Marcotte, the chief engineer for the Water and Sewer Authority, said no one from Givoo ever approached him with safety concerns.

The company, he added, had an obligation to continue bringing any safety shortcomings to the city's attention.

"Certainly I would like to hear from our contractors and our people if there is an issue of health and safety of that magnitude," Marcotte said. "I am very troubled by their take on this."

Marcotte said yesterday that the authority is installing new chlorine sensors and intends to spend $11,000 to replace its chlorine monitoring system. Yesterday, the authority suspended consideration of Givoo's contract renewal.

Chlorine Overcomes Men

Even with weaknesses in Blue Plains's chlorine warning system, a serious accident that causes widespread casualties is unlikely, said Gardner Bates, a spokesman for the Chlorine Institute.

Chlorine, he noted, was safely used in plants across the nation before many modern alarm systems were invented.

But Bates agreed that without properly working safety monitors, the likelihood of small releases inside a treatment plant could be raised, because workers might take longer to determine that something is wrong.

Four times in the past five years, at least trace amounts of chlorine have been released improperly at Blue Plains. In none of the cases was the public immediately notified.

On Aug. 15, 1994, a storage tank that workers thought was empty suddenly sent a plume of chlorine into the air. Four men in two boats were nearby that night, fishing on the Potomac.

Two of the men described the incident this way: A puff of what they thought was steam became a white cloud, which soon surrounded them in a choking fog. They couldn't breathe; they couldn't scream. They pulled their shirts over their heads for protection, to no avail. Their eyes watered and their skin itched.

Mark Trodden, of Springfield, one of fishermen, recoils at the thought of that night.

"It was an incredibly horrible night," he said. "It is a chapter in my life I want to get over with."

All four men were admitted to hospitals, and one says he and his fishing partner nearly died in the incident. The city and the EPA investigated; the incident ultimately was blamed on a pressure gauge that had broken off the tank.

The other recent releases were small enough that they did not require a formal probe.

In May, a part in the plant's chlorine feeding system ruptured, sending a plume of less than 10 pounds' worth of chlorine into the air, said D.C. officials, who provided no other details on the incident.

In June, the chlorine detection alarm went off when no one was at Chlorine Building I.

In August, a Givoo technician was checking the chlorine building when the alarm went off. He reset it, but within seconds, it sounded again. Employees soon noticed the smell of chlorine, and the plant safety office was alerted, says an Aug. 20 memorandum on the incident.

No one was reported injured, but with the third release since spring, several technicians said they began to seriously question the plant's safety.

"They have got a problem there," said one Givoo technician, who agreed to be interviewed because he was afraid a serious accident could happen at Blue Plains.

"If this keeps one man breathing so he can go home and see his family," he said, "I guess it is just worth it."

Staff researcher Bobbye Pratt contributed to this report.

RISK AND SAFEGUARDS

Chlorine at Blue Plains

Blue Plains Wastewater Treatment Plant processes more than 300 million gallons of wastewater a day, using more than 1,300 gallons of liquid chlorine to help sanitize the water before it is released into the Potomac River. Two-thirds of this chlorine is used in Chlorine Building I.

If released into the atmosphere, chlorine quickly turns to a gas, expanding 460 times in size. In an area immediately around a spill, chlorine gas can measure more than 20,000 parts per million.

Chlorine exposure threshold (in parts per million)

0.2 to 0.4 ppm Odor perception

1 to 3 ppm Mild mucous membrane irritation

5 to 15 ppm Moderate irritation of the respiratory tract

30 ppm Immediate chest pain, vomiting, coughing

40 to 60 ppm Lung tissue swells with fluid

430 ppm Lethal over 30 minutes

1,000 ppm Fatal within a few minutes

Potential danger

Chlorine is shipped to Blue Plains in 90-ton rail cars. A catastophic rupture of a chlorine rail car is unlikely. Such a rupture could place people within a 3.4-mile radius at risk of death and might cause non-lethal but harmful reactions in people as far as 31 miles away.

Even a small-scale chlorine release could pose a large threat. In a scenario outlined by the plant, a 10-minute release of chlorine from a hose leak could create a life-threatening plume that would spread over a third of a mile, placing plant workers and those at the Naval Research Laboratory and in Navy housing at immediate risk.

Detecting chlorine

Chlorine Building I is equipped with seven devices designed to detect even trace elements of chlorine in the air. The sensors must be tested periodically -- usually by private contractors hired by the Water and Sewer Authority. But documented plant practices have deviated from the recommended testing procedures, yielding what may be unreliable results. The two tests detailed below are vital:

1. Quick check (twice a month)

Fast method of testing the sensors for response to chlorine

Recommended procedure: A plastic cup is filled with a small amount of calcium hypochlorite (a dry form of chlorine) and held beneath the sensor.

Response time: Five seconds

Documented plant practice:

Procedure: A

potent solution of bleach and vinegar is used to trigger the sensor.

Response time: Despite the inten-sity of the solution, records show that some sensors failed to respond after five minutes.

2. Calibration procedure (semiannually)

A more accurate method of ensuring the device is adequately sensitive to chlorine.

Recommended procedure: Sensor is exposed to a known concentration of chlorine gas from a calibrated sample.

Response time: Five minutes

Plant practice

Procedure: In their most recent attempt to conduct this test, contractors said that they were given a calibrated gas sample that had expired and that they later lacked a formula needed to conduct the test.

The sensors

Chlorine gas sensors have an expected life span of 18 to 24 months. Records show a wide disparity in the ages of the seven sensors in Chlorine Building I, with at least one sensor manufactured more than nine years ago.

SensorLocationAge (From manufacture date)

Sensor 1Near rail car7 months

Sensor 2Near rail car9 years, 9 months

Sensor 3Basement3 years, 9 months

Sensor 4Basement 5 years, 11 months

Sensor 5Basement11 months

Sensor 6First floor3 years, 8 months

Sensor 7First floor5 years, 11 months

SOURCES: Blue Plains Wastewater Treatment Plant, the Chlorine Institute, District of Columbia Water and Sewer Authority, Scott/Bacharach Gas Detection Products