Polymetallurgical Corp. - Executive Summary |
EXECUTIVE SUMMARY RISK MANAGEMENT PLAN POLYMETALLURGICAL CORP. ABOUT POLYMETALLURGICAL Polymetallurgical, a wholly owned subsidiary of Cookson Corporation, is one company helping the Attleboro area retain its reputation as the global center for the manufacture of high quality metallic composite strip and wire. These unique mill products are used in the manufacture of electrical connectors, thermostatic controls, electronic assemblies, automobiles, and a broad array of high technology products. Started in North Attleboro in 1972 and located at 242 Broad Street since 1990, Polymetallurgical (PMC) currently employs 75 full-time people. It is responsible for over $1 million per year in taxes and wages paid within the town of North Attleboro and immediately adjacent communities. Sensitive to its location very close to residential areas and on the shore of Whiting Pond, the company has taken pains to minimize noise, air emissions, and wastewater pollu tants emanating from the facility. This executive summary describes recent efforts to evaluate and to further minimize safety risks associated with the use of anhydrous ammonia at 242 Broad Street. ABOUT THE USE OF AMMONIA In order to successfully form and shape its metallic products, PMC -- like many other companies in the area -- uses pure, dry (anhydrous) ammonia to create a protective atmosphere of nitrogen and hydrogen within special metalworking furnaces. The ammonia is delivered to the facility by tank truck, roughly once every 8-12 weeks. It is received and stored as an unrefrigerated liquid kept under pressure. The ammonia storage tank is a specially designed, aboveground, steel cylinder of 8000-gallon capacity. Industry-standard procedures and special design features limit filling to 85% of the maximum tank capacity. The pressure inside the tank rises in summer and falls in winter as the temperature of the liquid ammonia changes inside the tank. In order to ensure that the pressure inside the tank is sufficient to deliver ammonia gas inside the plant, the tank is kept warm in winter by a small electric heater. Ammonia gas is piped from the tank, through an emergency shut-off valve and several pressure regulators, to electrically heated dissociators located on the roof of the building. There the ammonia is 'cracked' at a pressure of 15 psig into a mixture of 3 parts hydrogen gas to 1 part nitrogen gas. The cracked gas contains virtually no ammonia, and the unused hydrogen is safely burned as it escapes from the entrance and exit of each metalworking furnace. GENERAL PRECAUTIONS AGAINST THE ACCIDENTAL RELEASE OF AMMONIA Several national standards organizations -- including the Compressed Gas Association (CGA), the American Society of Mechanical Engineers (ASME), and the American National Standards Institute (ANSI) -- have long recognized the possibility of the accidental release of various fuels and chemicals from pressurized storage tanks, and they have developed detailed design requirements and operating guidelines to prevent such accidental releases. The ammonia storage tank was designed and installed in full compliance with these standards and guidelines, and it incorporates such safeguards as: * the exclusive use of heavy-gauge steel pipe for all permanent connections between the tank, the building, and the truck unloading station, designed to prevent accidental breakage * restriction orifices and excess flow check valves on all connections to the tank, designed to minimize or stop an accidental release if human error causes a pipe to be opened or disconnected at the wrong time * hydrotesting of the tank to prove it can withstand an internal pressure far in excess of its design pressure * relief valves, designed to spit very tiny amounts of liquid ammonia out of connecting pipes as necessary to prevent a more serious and irreversible pipe break if cold liquid ammonia is trapped w ithin a section of pipe by human error * gas release valves, designed to keep the tank under its design pressure if the tank is exposed to a nearby fire * exclusion of any combustible materials from the vicinity of the storage tank * enclosure of the tank by a chain-link fence topped with barbed wire, monitored by television cameras and intrusion detection alarms, equipped with nighttime lighting, and located well inside the plant fence line. Standard operating procedures require that the driver of the ammonia delivery truck be specially trained and equipped to prevent spills during unloading and to deal with spills in the unlikely event that they occur. Procedures also require that a PMC employee be stationed at the ammonia storage tank during unloading, to make sure that the driver follows safe procedures and to monitor the level and pressure inside the tank. ACCIDENT HISTORY In the nine years that Polymetallurgical has owned the current facility, and over the 27 ye ars that it has operated in North Attleboro, the company has experienced no accidental releases from the ammonia process that resulted in deaths, injuries, or significant property damage onsite, or offsite. Nor have any evacuations, sheltering in place, property damage, or environmental damage occurred. In 1998 there was a very small leak of ammonia from a pipe fitting inside the building, close to the storage tank. PMC notified the fire department (NAFD) of this occurrence and had the department stand by during repair, just in case any emergency response was required. The leak was repaired without incident. WORST-CASE RELEASE SCENARIO Despite the safety provisions and perfect safety record described above, EPA regulations require consideration of a theoretical worst-case release scenario -- the failure of the largest single vessel containing ammonia. PMC's worst-case scenario is failure of the 8,000 gallon storage tank when completely full (85% capacity), such that t he entire contents empty over a ten minute period. EPA's Off- Site Consequence Analysis Guidance suggests a very conservative (i.e. very high) estimate of the distance to the endpoint (i.e., the distance that the ammonia gas may travel until it dissipates to a certain, prescribed concentration). The crude model offered by the agency for this purpose suggests an impact zone stretching 2.3 miles from the plant, potentially in any direction. The model assumes conditions that maximize the concentration of the gas and minimize natural dilution; such as assuming the entire contents of the gas are released within 10 minutes and that there is minimal wind or atmospheric turbulence (i.e. 3.4 mph and F-stability). Compared to other models, this model is known to do a inferior job of predicting the behavior of ammonia releases. To obtain a more realistic picture of the worst-case release scenario, PMC turned to the SLAB model developed by Lawrence Livermore National Laboratory, valid ated against large experimental releases of pressurized ammonia liquid and gas in the Nevada desert in 1986, and endorsed by the EPA as an acceptable model for risk management planning. Using the same unusually low wind speed and atmospheric stability assumptions, and given the same release rate 3500 pounds per minute (lb/min), this model predicts a distance of 1.24 miles to that same concentration endpoint. The concentration endpoint specified by the Agency is 200 ppm (parts per million by volume), the maximum to which nearly all individuals could be exposed for up to 1 hour without experiencing serious health effects or developing symptoms which could impair their ability to take protective action. In fact, nearly all individuals could be exposed for up to 1 hour at 1000 ppm without experiencing or developing life-threatening health effects; the maximum distance at which that high concentration could be encountered is predicted by the SLAB model to be 0.33 miles. Using the local average wind speed of 10 mph, and considering that 62% of the time the local atmospheric stability is Class D, the SLAB model predicts a still smaller impact zone as summarized here: Worst Case Release Scenario Wind speed/Stability 3 mph, F 10 mph, D Distance to 200 ppm 1.24 miles 0.36 miles Distance to 1000 ppm 0.33 miles 0.11 miles ALTERNATIVE-CASE RELEASE SCENARIO - WITHOUT MITIGATION The alternative release scenario is considered to be much more likely to occur than the worst-case scenario, but still very unlikely to occur in practice. Essentially, in requiring consideration of an alternative case, the EPA is saying, "OK, we know that the worst-case scenario might be unimaginable, but think about what could conceivably go wrong here." Accordingly, PMC has imagined that the connection between the ammonia tank truck and the storage tank is somehow completely broken during unloading, most likely by human error in making the connection. Now both the storage tank and the tank truck are equipped with excess flow check valves (EFCV), and these devices should immediately shut off the release from both locations. But for planning purposes, we assume that the break occurs in such a way that the EFCV on the storage tank fails to close, permitting liquid ammonia to escape at 225 gpm from the fill line that is connected to the bottom of the tank. This release continues at ground level until the tank is empty or until it has been shut off by the NAFD. Approximately 23% of this liquid immediately turns to a gas as it escapes from the tank and is no longer kept under pressure. The evaporation of liquid ammonia drops the temperature of the evaporated gas and remaining liquid to -28 F, disperses almost all of the remaining liquid as a fine mist, and condenses moisture in the air to form a thick fog. The liquid ammonia quickly evaporates as the air and ground warm it. And the temperature of the ammonia gas rises quickly as it mixes with the much warmer atmosphere and as the ammonia reacts with the water fog in the air. With a molecular weight of 17, ammonia gas is 41% lighter than air so, as the cold, foggy cloud of ammonia warms, it gradually rises above the ground. The SLAB model predicts that the ground-level concentration of ammonia from this alternative release scenario will be: Alternate Release Scenario - Without Mitigation Wind speed/Stability 3 mph, F 10 mph, D Distance to 200 ppm 0.40 miles 0.17 miles Distance to 1000 ppm 0.17 miles 0.07 miles MITIGATION OF ACCIDENTAL AMMONIA RELEASE RISK Although the likelihood of any accidental release is remote, PMC finds the impact zone predictions described above to be unacceptable, and is now (June, 1999) deciding what method or methods it will employ to reduce that risk. The options being considered include: 1) Reducing the maximum quantity of ammonia stored in the existing equipment 2) Replacing the existing 8000 gallon storage tank with two 1000 gallon tanks 3) Installing an emergency water curtain system capable of absorbing most of an accidental ammonia release before the vapor could drift offsite 4) Enclosing the storage tank and connecting that enclosure to a scrubber capable of absorbing most of an accidental ammonia release 5) Refrigerating the ammonia and installing an insulated dike to hold any release as liquid, minimizing the release of ammonia vapor 6) Eliminating the use of anhydrous ammonia, and instead generating hydrogen gas on site by the electrolysis of water or by reforming natural gas 7) Eliminating the use of anhydrous ammonia, and instead supplying the hydrogen needs of the metal treating furnaces from a cryogenic hydrogen tank The first option will be pursued as quickly as possible, with the objective being to reduce and then to maintain the inventory below 10,000 lb. PMC recognizes that both the first and second options will increase the probability of a release because of the increased frequency of deliveries that will be required to operate with a lower ammonia inventory; and the second option will approximately double the number and reduce the size of the fittings that might break. On the other hand, options 1 and 2 both decrease the impact zone of the worst-case scenario, and option 2 would decrease the impact zone of an accidental connection failure. Ammonia releases are easily and very effectively absorbed by water. Numerous large-scale field tests have demonstrated that water sprays can reduce the impact zone by 90% or more; but the quantity of water required may exceed the capacity of the city water system. Fixed scrubbers can be even more effective as long as the release point can be enclosed or otherwise captured into the scrubber; but scrubbing of a very cold, very fast release can be problematic, and capturing a release like that described as the 'alternate release scenario' might be difficult. Both options 3 and 4, if ever exercis ed, would generate sizable wastewater streams that would be quite difficult to manage on site. Option 5, refrigeration of the stored ammonia would greatly reduce the impact zones under any scenario. But it introduces new costs and complexity to the operation. Option 6, onsite hydrogen generation is the most complex and is likely to be the most expensive. The capital investment could easily be more than $500k; operating and maintenance costs are likely to be higher than those of other options. Although hydrogen is routinely generated for other applications by the methods being contemplated, the technologies are largely unproven in this application and size range. This option carries a higher technical risk of project failure, but promises near-zero inventories of flammable and toxic materials. The substitution of bulk hydrogen delivery for ammonia dissociation has already been adopted by a number of other metal forming companies in the area. It is a well-proven, commerci ally available option. However, cryogenic hydrogen storage would substitute a new hazard (deflagration/detonation) for the existing toxic release hazard, and the hazard zones for that new hazard have yet to be defined. In summary, there is no easy solution to the high-consequence/low-probability risk associated with anhydrous ammonia use at PMC. But the company is committed to reducing that risk, and it expects to eliminate the risk of an offsite toxic impact within the next twelve months. Comparison of Risk Reduction Measures: Distance to Endpoint, as Predicted by SLAB Model Worst Case Alternate Scenario No change, current inventory (35,000 lb maximum)---- 1.2 miles ---- 0.4 miles Reduced inventory (9,500 lb maximum) ---- 0.7 miles ---- 0.4 miles Small tank (850 lb maximum/tank) ---- 0.4 miles ---- 0.1 miles Current tank or small tank, with active mitigation---- TBD ---- < 250 feet Refrigerated ammonia storage ---- < 0.2 miles ---- < 0.2 miles Onsite hydrogen gene ration ---- NA ---- NA Onsite hydrogen storage ---- (No toxic risk. Deflagration/detonation risk TBD) (NA = not applicable. TBD = to be determined) EMERGENCY RESPONSE PROGRAM The PMC Emergency Management Plan calls for close cooperation with the NAFD in dealing with spills or releases of hazardous materials at the facility. If ammonia is kept on site, and if a release occurs during working hours, the NAFD would be notified immediately by security, the onsite emergency response coordinator would quickly determine if evacuation of the facility was required, and personnel would quickly move upwind, away from the release. Plant maintenance staff wearing protective equipment maintained on site could handle small gas leaks. Larger releases at the storage tank could be mitigated at present by the NAFD using fog nozzles, or in the future by other means described above. Outside of working hours, if an active mitigation system were installed, a significant ammonia release would trigger operation of the water curtain or scrubber and simultaneous alarms at the NAFD and company's security center. As a result of the recent risk management planning activity, PMC and NAFD have agreed to conduct regular joint training exercises to review and to practice their responses to a hypothetical ammonia release at this facility. PMC has suggested that an audible alarm could be installed, promptly warning residents in the immediate vicinity of the plant in the event of an ammonia release. PMC intends to work with the NAFD, providing technical advice and funding for a brochure that will inform local residents on the actions they should take if an ammonia release was ever to occur. In sum, the emergency response plan is a multi-layer defensive strategy: * First, the plant is designed so that it is extraordinarily unlikely that a significant ammonia release would ever occur * Second, as evidenced by its perfect safety record, the plant is maintained and opera ted in a very safe manner, consistent with the underlying design * Third, the local fire department is trained and equipped to respond promptly and effectively to any release of ammonia at the site. Future installation of a water curtain or scrubber system could provide a nearly instantaneous and even more effective response. Other measures now under consideration would diminish or eliminate the need for such active mitigation measures. * Fourth, unless an offsite impact can be completely avoided, the appropriate local authorities will inform the public of actions that should be taken if a significant release should ever occur. |