Army Has Hard Time Meeting Terms Of EPA Permit at Johnston Island

posted in: March 1997 | 0

Johnston Island lies about 800 miles southwest of Hawai’i. As a geologic feature, the islet is but a speck of coral that has been artificially expanded by United States military forces to 625 acres, nearly 15 times its original 42-acre size. As the site of the Johnston Atoll Chemical Agent Disposal System, however, Johnston Island occupies a pre-eminent place in the history of chemical weaponry.

JACADS is the first in what was to be a series of chemical weapons incinerators that the U.S. Army planned to build at nine Army installations around the country. Just one other facility has been built and is in operation at Tooele, Utah. Whether the remainder of the planned incinerators are ever completed or put into operation is in question. In the 1997 defense budget, Congress inserted a requirement that the Army review once more alternatives to incineration of chemical weapons. Should those alternatives prove to be effective and more acceptable to the communities near the bases where the weapons are stored, it is, of course, possible that no more incinerators will be built. For Johnston Island, any talk of an alternative chemical weapons disposal technology is probably academic. Still, a review of the operating history of JACADS may be useful in evaluating the very need for an alternative.

Operation Red Hat

Chemical weapons were first used in World War I. Relatively simple to manufacture, they are nonetheless difficult to eliminate. Until the 1960s, commonly used disposal techniques included burning them in open pits disposing of them by burial on land, or dumping them in the ocean. Burning at-sea on incinerator ships was tried for a time, but public outrage put an end to that option. (The last chemical munitions ocean dump occurred in August 1970.)

In 1970, Congress required the Department of Health and Human Services to review any chemical weapons disposal plans that the Department of Defense might develop. At the same time, it restricted the movement of chemical weapons.

All this created a problem for the Army. In the 1960’s, it had begun planning to remove its chemical weapons stockpiles from Okinawa. Congress forbade their return to the continental United States, so the Army decided to store them indefinitely at Johnston Island, in an operation known as “Red Hat.”

In 1984, the National Research Council reviewed disposal options for chemical weapons, with the two main technologies considered being incineration and chemical neutralization. The latter method, the NRC decided, was more costly and produced larger quantities of waste than incineration. Based on the NRC report, the Army developed plans to incinerate stockpile chemical weapons, and in 1986, it submitted its plan to do so to Congress. In 1988, the Army formally announced that on-site incineration was its preferred disposal method for stockpile chemical weapons, and selected Johnston Island as the site where it would demonstrate the efficiency and safety of its program to incinerate chemical weapons.

To be sure, Johnston Island was not the first site where such arms had been burned. It was, however, to be home to the first large-scale chemical weapon incinerator. Construction work was completed in 1989, and on June 30, 1990, JACADS incinerated its first chemical weapon a GB-filled M-55 rocket.

In late 1990, Johnston began receiving shipments of chemical weapons that had been part of the U.S. stockpile in Germany. A year later, the Army discovered 109 World War II-era mustard-gas-filled projectiles in the Solomon Islands. These, too, were shipped to Johnston.

At the time, destruction of chemical weapons began at JACADS, roughly 5.2 percent of the total U.S. stockpile of chemical weapons was on the site. When the U.S. stockpile in Germany was added to that, the figure increased to 6.6 percent.

Under an ambitious schedule, the Army planned for JACADS to complete destruction of the Johnston Island stockpiles by 1994. In January 1996, after five and a half years of operating, the Army announced that JACADS had destroyed about 2 million pounds of nerve gas and mustard agent, representing about a fourth of the total stockpile of agent stored on the island. It wasn’t easy.

Thousands of Glitches

It is only reasonable to expect glitches in any type of chemical weapons disposal process. JACADS has certainly had its share.

JACADS has four main incinerator chambers: the metal parts furnace (MPF), where the metal shells of rockets and ton containers are burned; the liquid incinerator (LIC), where agent drained from the shells is burned; the deactivation furnace (DFS), where explosives and propellants are destroyed; and the dunnage incinerator (DUN), intended to burn packing and other miscellaneous contaminated materials. All these are connected to an exhaust system intended to prevent releases of hazardous emissions.

The incinerators are designed to burn at such high temperatures that no dangerous residues from chemical agents remain on the burned material. A system of alarms is to be triggered if the furnace temperature drops below the minimum established. The alarms are supposed to be tied in to a stop-feed mechanism, so that the whole incineration process comes to a halt if the temperatures fall below those minima.

Another system of alarms is to be triggered if the concentration of chemical agent in the emissions stack exceeds a certain level (0.0003 milligram per cubic meter for GB or VX; 0.03 mg/cubic meter for mustard gas). Once again, the alarms are supposed to cause the system to shut down, ensuring that any problem will be corrected before a significant release of gas occurs.

Despite the precautions, there have been problems. Each year, the Army is required, under terms of the RCRA permit, to report all instances of noncompliance, whether or not they involve releases that could harm human health or the environment. Copies of the 1990 through 1995 reports were obtained by Environment Hawai’i. They disclose literally thousands of instances of non-compliance. Some are as relatively insignificant as failure to keep proper records or missed inspection or maintenance schedules.

Others are more serious, even if they are not associated with environmental harm. One of the earliest non-compliance events occurred in 1990, when computer files that contained a record of certain monitoring data for test burns were erased.

The 1991 report notes that, among other things, alarms indicating when hazardous storage waste tanks were near overflow conditions were manually disabled during JACADS operations, as were the associated controls that would automatically shut off flows to the full tanks. Workers apparently found it easier to have the waste spill over to the secondary containment areas rather than stop operations until adequate primary storage was arranged.

A similar problem was noted in the Army’s noncompliance report for 1992. The temperature in the liquid incinerator dropped below the minimum of 2550 degrees Fahrenheit, but the automatic control that was supposed to stop operations was not triggered. As the Army reported to the EPA, its investigation determined that the automatic shut-off control had been disabled seven weeks earlier.

These themes and variations continue throughout the Army non-compliance reports for later years as the Army confesses, in mind-numbing, eye-glazing detail, to literally thousands of infractions of the tens of thousands of operating conditions set forth in the EPA permit.

… And Some Whoppers

Sources other than the Army reports provide a better idea of how JACADS has or hasn’t worked. The General Accounting Office of the U.S. Congress has published numerous reports on JACADS in particular and the Army’s chemical weapons disposal program in general. In 1993, the GAO wrote:

“Destruction operations at Johnston Island were slowed due to unexpected maintenance problems, which occurred on an almost daily basis. During the last phase of the VX rocket campaign, when the prototype i.e., JACADS was operating at full rate, the Army demonstrated an equipment reliability rate of 5 percent, less than the campaign goal of 75 percent and considerably less than the 85 percent reliability rate predicted by the original equipment design contractor for normal processing of M-55 rockets. Because of these maintenance problems, the Johnston Island facility did not operate at all on 32 of the 105 scheduled processing days during the VX rocket campaign. Even after 44 weeks of both GB and VX M-55 rocket incineration experience, the prototype facility remained operational for the scheduled 10 hours less than 30 percent of the days during the full-rate (capacity) phase of the campaign.

“The greatest single source of maintenance downtime during the second campaign was caused by an explosion [January 21, 1992] in the deactivation furnace used to destroy explosive material. The system was shut down for 16 days to inspect and repair a 24 inch-by-8-inch hole in the furnace caused by the explosion.” The repairs were only temporary. In March 1993, an entirely new kiln, of more rugged design, was installed in the DFS.

The DUN System

The dunnage furnace has never worked more than a few hours in the entire history of JACADS. It was still in start-up mode, still awaiting its first load of contaminated dunnage, when an explosion – discreetly described by the EPA as a “pressure transient” – deformed the furnace chamber and building. In a report prepared by the EPA in July 1993, describing the events surrounding JACADS’ operational verification testing, the EPA went on to say: “After repairs and modification, shakedown with non-hazardous waste began in September 1992, and continued intermittently until March 1993, when, after a short shakedown with waste contaminated with HD, a DUN trial burn was attempted. This trial burn had to be abandoned after a few hours because of bypass damper control problems.”

Gary McCloskey, the on-site JACADS manager, told Environment Hawai`i that eventually the dunnage incinerator had undergone a trial burn of hazardous waste. But when asked if it was now working, he acknowledged it was not. In any case, the volume of material that was intended to be burned in the dunnage incinerator was turning out to be smaller than planners had anticipated it to be – “we haven’t had a leaker problem at Johnston Atoll,” he said. When small amounts of miscellaneous contaminated waste accumulated, he added, it was sent to the metal parts furnace. “We have a large quantity of uncontaminated dunnage,” he said, which “we’re disposing of commercially.”

Fundamental Flaws

Operational verification testing (OVT, as the Army calls it) lasted until March 1993. Some of the test burns were observed by EPA staff. The Army itself contracted with the MITRE Corporation to prepare reports on JACADS’ trial burns.

Environment Hawai’i has not been able to locate copies of the book length MITRE reports. However, Pat Costner, a chemist who consults to Greenpeace, was able to them. In a 1993 paper on the trial burns of nerve gas (GB and VX), Costner stated:

“The demilitarization of chemical weapons at JACADS has been accompanied by myriad problems, ranging from basic design flaws to repeated mechanical failures during extensive tests with rockets filled with GB as well as those filled with VX. Some, but not all, of these problems have been remedied.

“Detoxification/disposal by incineration has suffered from an even broader range of problems. Many of these problems are inherent limitations of this technology, and, consequently, essentially unresolvable. For example, incinerators simply do not achieve the requisite destruction and removal efficiencies on substances that are present at relatively low concentrations. Further, given the basic design of JACADS, malfunctions during the separation process contribute to incinerator upsets which are, in turn, associated with increased emissions of unburned agent and products of incomplete combustion.”1

Limited Efficiency

Costner concluded that the analytical systems used to detect unburned chemical agent “had unacceptably high rates of malfunction during both the GB and VX campaigns…. During the GB campaign, the Army failed numerous times in its efforts to isolate active chemical agent from the work environment and public domain: on 32 occasions, active agent was released into the corridors frequented by workers; on 15 occasions, active agent was detected in the life support air system; on five occasions, identified as ‘likely’ false positives, active agent was detected by perimeter monitors; on at least one occasion, active agent was evidently released from the incinerator stack, although the stack monitor was not functional during this event.” Similar problems arose during the VX campaign.

One of the central problems of JACADS, alluded to by Costner in the quotes above, is the difficulty of achieving targeted destruction and removal efficiencies (DRE) when dealing with substances present at very low concentrations. JACADS has a DRE target of at least 99.9999 percent. However, Costner cites studies indicating that when concentrations are less than 1 percent by weight (10,000 parts per million), this rate of destruction is unachievable. As the concentrations fall even lower, so, to, do removal efficiency rates. “For those chemicals present in wastes at concentrations of 100 ppm or less,” Costner writes, “no incinerator was able to achieve a DRE of 99.99 percent. Yet these very low concentrations are typical of agent levels present in wastes burned by the deactivation furnace, the metal parts furnace, and the dunnage incinerator (on the few occasions it has worked), Costner notes. “Further,” she continues, “even if the Army’s incineration system were able to achieve a DRE of 99.99999 percent, as reported for BG in the LIC [liquid incinerator], or 99.999999 percent, as reported for VX in the LIC, during every moment of operation with all chemical agents fed into each of the incinerators, the quantities of unburned agents released in stack emissions are sufficient cause for concern for public health and the environment.”

In addition to releasing unburned agent, the stacks at JACADS also release products of incomplete combustion, including dioxins. Yet, Costner writes, “the issue of products of incomplete combustion was not an active area of concern” during the JACADS trial burns.

Down Time

During the VX and the GB campaigns, JACADS was down more than it was operational. The GB campaign lasted from July 16, 1990, through February 27, 1991. During that time, “JACADS accomplished 500 hours of active demilitarization and incineration, accompanied by a cumulative downtime of 929 hours.” During that same 500 hours of operation, “the network of monitors for detecting GB releases triggered 776 major process alarms,” Costner writes, “an average of 22 per day.” MITRE attributed most of these to high levels of carbon monoxide, but Costner notes that carbon monoxide levels are “commonly used as a surrogate indicator of incinerator performance, because high CO levels increase during major upset conditions, which are also accompanied by high PIC [products of incomplete combustion] emissions.”

The down time experienced during the VX campaign was even greater. According to the MITRE report, Costner writes, “472.8 hours of actual rocket processing were accompanied by 1496.4 hours of downtime. i.e., every hour in which JACADS was processing rockets was accompanied by even more than three hours of downtime.” There were 3,691 major process alarms – “more than 7 alarms per operating hour,” Costner observes – during the VX testing.

Dipsticks

In 1994, a year after her report on the VX and GB campaigns, Costner and Greenpeace released a report on the trial burns of mustard gas (HD) in the metal parts furnace and the demonstration burn of UD in the liquid incinerator. “This report,” Costner writes, “explores in detail the various ways in which the 1992 trial burn with agent HD in the metal parts furnace (MPF) and the demonstration burn with HD in the liquid incinerator (LIC) failed to meet legal criteria, federal guidelines, and common standards of practice in incinerator operation and chemical analysis.”2

At the heart of this list of problems is the fact that over time, the chemical agent inside some of the bulk containers stored at Johnston Island has been transformed into something other than what it started out as. Under ideal conditions, the projectiles containing chemical agents or bulk containers (called ton containers, although the weight of the agent in them is undetermined) are completely drained before the metal shells are burned in the metal parts furnace and the drained liquid is sent to the liquid incinerator.

In the case of the ton containers treated in the MPF trial burn and the LIC demonstration burn, a substantial portion of the agent had gelled and could not be drained out of the containers. The RCRA permit anticipated that at least 95 percent of the agent would be removed from the containers by draining, and that no more than a 5 percent residue of agent would remain on the metal parts sent to the MPF. However, the chemical transformation of HD into a solid or semi-solid prevented the draining of anywhere near this percentage of material from the metal shell.

The chemical change meant that the Army no longer had any clear idea of exactly what it was sending to the incinerators, or even of the quantities of substance burned. “Indeed,” Costner writes, “the procedures followed in the Army’s attempt to determine the feed-rate of HD and its accompanying materials to the MPF during the trial burn are remarkable for their creativity if not their accuracy.

“Workers pushed wooden dipsticks through holes until they reached the bottoms of the containers and noted, on the dipsticks, the approximate depth of the materials in the containers. Then, in some undisclosed fashion, the marks on the dip-sticks were related to the volume of undrained material in each TC [ton container], which was, in turn, related to the density of this material, which was, in yet another turn, determined or calculated by another undisclosed method.”

This method of calculation assumed an even distribution of residual heel in the ton containers. However, Costner notes, the material was, in some containers, unevenly distributed.

Discrepancies

Costner told Environment Hawai`i that it was difficult to believe that the Army could claim its operational verification tests were successful on the basis of the MITRE reports. She is not the only critic. The U.S. General Accounting Office has also voiced skepticism about the results of the OVTs.

“We believe the methodology used by the Army and its independent oversight contractor [MITRE] overstated destruction rates achieved during the second operational verification campaign,” the GAO wrote in the 1993 report cited earlier. “For example, the contractor’s August 1992 draft assessment report cites and average overall destruction rate of 19.6 rockets per hour for the second campaign. However, the methodology the contractor used to compute the rate is not consistent with the initial test plan or the methodology used to calculate destruction rates for the first verification campaign. Using this initial methodology, we determined that the average hourly overall destruction rate was 12.2 rockets per hour” barely half the goal of 24 per hour that the Army had established for the tests. (For mainland facilities, the Army has established a goal of 32 rockets per hour. The JACADS goal was set below this to compensate for problems associated with operating in a remote location.)

The GAO also took note of the untimeliness of the MITRE reports: “The contractor was expected to publish separate evaluation reports suitable for public distribution within 90 days of completion of each operational verification campaign.” The first report on the GB campaign was on time. However, the report on the second campaign was not available until eight months after the test concluded. “Army officials told us they commented on several draft versions of the report,” the GAO wrote, “but final publication was delayed in reaching agreement on the Army’s final comments to the report.

Rewriting Permit Terms

A further complication resulting from the solidification of agent is that the permitted limit for burning of agent in the metal parts furnace has had to be routinely exceeded. This led to the Army requesting a change in its permitted operating conditions to allow for up to a 50 percent “residual heel” of agent in material that is sent to the metal parts furnace.

The use of incineration as a technique to destroy chemical agent has been justified on grounds that the very high temperatures in the liquid incinerator will destroy almost all of the agent. At lower temperatures, such as those allowed in the MPF, similarly high efficiency rates would not be expected.

In 1996, the Army proposed two major changes in the operating conditions of the metal parts furnace. First, it sought to have the condition limiting agent residue to 5 percent on material burned in the MPF lifted to 50 percent. Second, despite seeking an increase in the amount of agent residue sent to the MPF, it sought to have the minimum temperatures allowed in the MPF lowered still further – to 1300 degrees Fahrenheit, plus or minus 150 degrees, from the previous level of 1450 degrees, plus or minus 150.

In point of fact, this is how the Army has been operating its MPF, despite permit conditions. Costner writes that in the MPF trial burn of UD agent, “USEPA gave permission to expand the [operational] limit to 1450 plus or minus 250 F during the trial burn. Further, the OVT [operational verification testing] states that USEPA has given permission to expand the temperature limit further to 1450 1-300 after the trial burn.

“Each of these relaxations is contrary to normal practice. In particular, USEPA’s agreement to allow even less stringent control of the MPF temperature after the trial burn is contrary to the purpose of conduct mg a trial burn…. Moreover, each of these relaxations of the MPF furnace temperature limits increases the probability of poorer DREs and higher emissions of PICs.”

In any case, Costner notes, “the Army has operated the MPF at temperatures that are lower even than the most relaxed limits cited above. As described in MITRE’s OVT report, at least one container was processed at 1000 F, some is 150 F cooler than the lowest limit USEPA has allegedly approved.”

To justify the permit modification, the Army has said that the lower MPF temperature range is needed “to demonstrate the capability of the MPF to effectively treat hazardous waste at a feed rate higher than currently permitted.” The higher feed rate is needed if JACADS is to complete its mission on deadline – which is now set at the year 2004.

Costner is suspicious of these changes. In a telephone interview with Environment Hawai’i, Costner described the MPF as the “Achilles’ heel” of JACADS. “When you put metal in the MPF, it becomes a massive heat sink. You get a very uneven burn, and the temperatures in the chamber are difficult to control.”

Violations

Costner regards the EPA as having been extraordinarily lenient with respect to enforcement of JACADS permit terms or interpretation of their violations. Still, the EPA has issued several notices of violation to JACADS.

The most significant notice of violation was issued on March 13, 1995. At that time, the EPA notified the Army of three counts against it, for which the EPA was seeking payment of fines totaling $122,300. The first count involved release of 11.6 milligrams of GB to the environment on March 23, 1994.

(In fact, the actual release of GB was almost certainly many times this amount. An EPA inspection report dated December 1994 contains a description of the circumstances surrounding the release: “At approximately 10:57 PM on March 23, 1994, the incinerator stack ACAMS [Automatic Continuous Monitoring System] went into a high level alarm, registering a concentration of 0.003 mg/m3 of BG agent, which is equivalent to 10 times … the ASC [Allowable Stack Concentration],” and remained above this level for at least ten minutes. The recorded concentration, however, “is the maximum value of GB agent concentration recordable by the PDAR [Process Data Acquisition] system. Therefore, although the [Army] estimated that 11.6 mg of GB agent had been released, the actual amount may have been greater.”)

The EPA defined this as a major violation and proposed a fine of $50,000. “[I]n failing to follow standard operating procedures for purging the agent feed line, the facility’s actions resulted in an actual release of GB chemical agent – which can only be construed as substantial non-compliance with the regulatory requirement.”

Count two involved inadequate aisle space in the permitted hazardous waste storage area. For this “moderate/minor” infraction, the EPA proposed a fine of $4,000. Count three involved 148 days of storing hazardous waste in an area not specified in the permit. For this “moderated major” violation, a fine of $68,300 was sought.

Following negotiations and corrective action with respect to the aisle space, the Army and EPA agreed to a consent agreement and order. The Army agreed to change the operation of its liquid incinerator in a manner intended to prevent further releases of nerve agent and, to address the improper waste storage violation, it agreed to apply for formal permission to use the area, known as Hama Point, for storage of hazardous waste. (This permit was granted on February 18, 1997.) Finally, the Army agreed to pay fines totaling $91,700.

Permit Renewal

The Environmental Protection Agency issued a Recourse Conservation and Recovery Act permit to the Army for its JACADS installation on August 30, 1985. The permit set conditions for the first 10 years of JACADS’ operations.

JACADS’ original operating permit anticipated that the weapons destruction program would be completed by 1995. However, the Army now anticipates the task won’t be finished until nearly 10 years beyond that original deadline.

To cover its operations after the original permit expired, the Army applied for a new permit from the EPA in 1995. Following agency review, the EPA is nearing the final stages of granting the new permit. According to a spokeswoman for the EPA, the public will be able to review the draft permit and comment on it at a public hearing in Honolulu before final EPA action. No date has yet been set for the hearing, but notice of it will be published in the legal advertisement section of the newspapers, a copy of the cited AG opinion. In a letter dated April 15, 1996, Matsubara was informed that the AG’s office was “unable to locate, if one exists, the attorney general’s opinion referred to by Mr. Roger Evans during the October 27, 1995 hearing. Warren Price, then Attorney General, submitted a letter to the county of Kaua’i regarding the Special Management Area in Hanalei. We have enclosed a copy for your reference. However, Mr. Evans stated that this was not the opinion he was referring to.”

In his opening brief filed with the court in August 1996, Matsubara wrote: “It is undisputed that the board relied upon a nonexistent AG’s opinion and related misinformation presented by Mr. Evans in reaching its decision on appellants’ petition. Further, appellants had no prior notice that this AG’s opinion would be raised at the hearing thereby precluding them from raising any meaningful defense on their behalf. Clearly, appellants were the victims of the very types of governmental actions the due process clauses were meant to prohibit.”

Against this, the best argument the DLNR’s attorney could raise was the claim that, “There is no evidence that the Land Board relied upon the alleged AG’s opinion or entertained any SMA-related consideration in making its decision.”

The Verdict

The missing or non-existent AG opinion was, among all the complaints raised by Matsubara, the one that appears to have resonated with Judge Chang. In his ruling and order filed on January 2, 1997, he determined that “Appellants’ constitutional rights to due process have been violated in this case.” For that reason, he went on to say, “It is not necessary for the court to address the remaining issues regarding whether the board exceeded its statutory authority, whether the board should be stopped from relying upon certain rules and statutes, and whether the board’s denial of appellants’ petition constitutes a taking of appellants’ property.”

Chang clearly rejected the attorney general’s argument that the missing AG opinion played no role in the board’s decision on the time extension request, citing the “circumstances preceding and surrounding Mr. Evans’ post-executive session presentation at the October 27, 1995, Land Board hearing” and the fact that the December 13, 1995, letter to the Engelstads “tracks the purposed subject matter of the non-existent AG’s opinion” in stating that no permits may be issued without SMA clearance.

The judge’s ruling does not, however, exempt the Engelstads from having to comply with SMA requirements. At present, two court challenges to the city’s imposition of SMA requirements on the Engelstads are pending.

Nor does it exempt them from having to comply with such DLNR rules and regulations relating to construction in the Conservation District. The last paragraph of Judge Chang’s order states, “Appellants Ralph L. Engelstad and Betty N. Engelstad must comply with all orders and decisions issued by the Court or any governmental entity and all rules, regulations, and the statutes applicable to the construction of the residence.”

(For additional background, readers may consult articles in the following editions of Environment Hawai`i: September 1990, [url=/members_archives/archives_more.php?id=532_0_33_0_C]February 1991[/url], [url=/members_archives/archives_more.php?id=813_0_32_0_C]April 1992[/url], and [url=/members_archives/archives_more.php?id=1148_0_29_0_C]December 1995[/url].)

1. Pat Costner, “Chemical Weapons Demilitarization and Disposal: Johnston Atoll Chemical Disposal System, GB and VX Campaigns,” released by Greenpeace (Washington D.C.), May 12, 1993.
2. Costner, “The Incineration of HD Agent at JACADS: MPF Trial Burn and LIC Demonstration Burn,” Greenpeace (Washington, D.C.), March 17, 1994.

Volume 7, Number 9 March 1997

Leave a Reply

Your email address will not be published. Required fields are marked *