Lessons Learned from an Anhydrous Ammonia Release Incident

Author: John J. Pallen; J.R. Simplot Co., Lathrop, CA 95330 (initial publish by AICHE Safety in Ammonia Plants and Related Facilities Symposium in 1989)


An anhydrous ammonia leak from the base of a 6,500-metric-ton atmospheric storage tank occurred at a California fertilizer manufacturing facility in March 1986. Lessons learned from the mishap can help prevent such incidents and their negative public impact.

A leaking anhydrous ammonia atmospheric storage _tank at the J. R. Simplot Company fertilizer manufacturing facility in Lathrop, California, attracted· much public attention for three days in March of 1986. The leak tarnished the good reputation of the company within the local community and created disagreement between public jurisdictions. Even after three years, the ammonia leak is still mentioned as a footnote in many local newspaper articles dealing with chemical releases.

In the wake of the incident, the company needed to ask some hard questions: Could the incident have been prevented? How well was the chemical release controlled? How well did the company communicate with emergency officials? Could the company have minimized the negative impact of the event? The answer to these questions may perhaps prevent a similar incident from happening again.

The incident

It began on Tuesday morning, March 18, 1986. At 4:30 A.M., the shift supervisor noticed a strong ammonia odor in the storage tank area. Upon investigation, a small leak of ammonia vapor was discovered coming from underneath the north shell of a 6,500 metric ton anhydrous ammonia storage tank. Because of the hazardous nature of ammonia vapor, the situation required prompt evaluation. If the leakage rate remained stable or decreased, only local drift control measures would be needed to prevent vapors from crossing the facility fence line; however, if the leakage rate increased dramatically or the tank ruptured, then more extreme actions would be needed to minimize vaporization and to control off-site drift.

Based on the evaluation, immediate action was taken to control fugitive vapors with water spray and to make arrangements for transfer of the anhydrous ammonia from the tank. Preparations to control a potential massive release from the tank were started.

Emergency notification procedures were put into effect at once upon detection of the leak. The incident ended on Thursday when all of the ammonia had been transferred out of the tank. The standby evacuation alert was called off, and Simplot workers began the two week process of cleaning and airing out the tank to allow for safe entry.

The cause

During the incident, the cause of the leak had been unknown. Simplot managers were concerned about why the tank had failed, and local officials wanted assurance that the tank was safe before being refilled. An investi­gation was launched to determine the cause of failure.

The tank stored liquid anhydrous ammonia at atmos­pheric pressure and ·at a constant temperature of -33°C (-28°F) from 1958 to March of 1986. It was 33.5 m (110 ft.) in diameter and 11 m (36 ft.) in height. The tank was fabricated by the Pittsburgh-Des Moines Steel Company and was built to ASME code, Section VIII, Paragraph UW-52. The floor plate material was specified to be ASTM A212 Grade B flange quality, fine grained practice normalized to A300 Class 1 to meet the charpy impact test of -45°C (-50°F). The 6.35 mm(¼ inch) thick floor plates were welded together using a square butt weld with a 3 .97 mm (5/32 inch) root opening on top of a 50. 8 mm (2 inch) wide by 6.35 mm (¼ inch) thick backup bar. The specified electrode for the floor welds was E7018, a low hydrogen electrode.

Examination of the tank interior was done on April 16, 1986, by Testing Engineers, Inc. Upon entering the tank, the investigators observed buckling of the floor in many locations·, and when walked upon, the buckled floor could be depressed. Corrosion inside the tank was minima] to nonexistent. A large crack 28 cm (11 inches) long was discovered approximately 60 cm (2 ft.) from the tank wall, oriented perpendicular to a weld and extending into the parent plates on both sides of the weld (Figures 1, 2 & 3.) Chevron U.S.A., Inc. engineers examined the tank the following week and discovered a smaller crack approximately 60 cm (2 ft.) from the tank wall, extend­ing 6.35 cm (2.5 inches) from a weld into the parent plate on one side only. The two plates containing the cracks were cut out and delivered to Testing Engineers’ laboratory in Oakland, California, on May 5, 1986.

Laboratory inspection of the backside of both plates revealed a 3 to 6 mm (1/8 to 1/4 inch) gap between adjacent backup bars on both the large crack and small crack plates (Figure 4). Also, there was a gap between the plate and backup bar on the small crack section. Examination of the fracture surface of both plates showed fatigue striations that originated in the center of the weld and propagated to the root (Figure 5). Microscopic examination of the weld did not reveal any weld defect.

Samples of the parent plate were tested for chemical and mechanical properties. Analysis for chemical composition indicated that the samples did meet the chemical requirements of ASTM A212 Grade B low temperature service. A mechanical analysis of the samples was done by testing for yield strength, percent elongation, and tensile strength. The samples did not meet the tensile requirements of ASTM A212 Grade B. Charpy vee notch samples were machined from both plate sections for impact testing. The parent metal did not meet the specified impact strength requirements.

The examination of the plate sections indicated that a fatigue crack developed and propagated in both welds until the applied stress was sufficient to cause brittle failure of the parent metal adjacent to the welds. Both cracks were perpendicular to the weld and did not have the backup bar support underneath them. It is most likely that cyclic loading on the tank bottom from changing levels of product initiated and propagated the cracks due to bending stresses locally intensified at the crack locations. The cracks may not have happened if contin­uous backup support had existed under the weld. Also, the deficient mechanical strength and impact resistance of the parent metal contributed to the initiation and propa­gation of the cracks. Improved impact resistance of the parent metal may have increased the life span of the tank.

The condition of the tank foundation was not inves­tigated. The lower foundation was constructed of a concrete ring footer wall filled with compacted sand and gravel layers. On top of this layer was a ring of light­weight concrete bricks filled with two layers of foamglass insulation. A waterproof membrane was placed on top of this structure. It is possible that some deterioration and subsidence of the foamglass insulation occurred, which could have contributed to buckling of the tank floor.

Control measures

At the beginning of the incident, the tank contained 2,000 metric tons of liquid anhydrous ammonia. The cold liquid was leaking at an estimated rate of less than 3 1/min. (1 gal./min.) and was quickly vaporizing upon contact with the underlying soil. Since the leak was discharging from the tank floor and not the side walls, rupture of the tank seemed unlikely; however, since the cause of the leak was unknown, action was commenced to control the small leak and to prepare for a potential larger release of ammonia.

The two immediate goals of controlling the leak were to minimize the amount of vapors leaving the site and to remove the contents of the failed tank. To control the vapors, two options were discussed. The first option was to fill the dike areas around the tank with water to absorb the leaking ammonia. This action could have caused thermal stress on the tank bottom and was considered too risky. The second option was to use water spray. Because of the small rate of leakage, the water spray was applied with fog nozzles directly to the outside periphery of the tank at the point of vapor emissions in order to completely solubilize the ammonia.

Problems were encountered using the fog nozzles. Shifting winds necessitated repositioning of the fog lines several times. During the first 24 hours of the incident, the winds changed from 29 km/hr (18 miles/hr) from the northwest to 8 km/hr (5 miles/hr) from the east. Swirling winds within the plant also made controlling the vapors more difficult. Another problem was the large quantity of water generated by use of the nozzles. The ammonia­bearing water was contained in the dike area and pumped continuously to a lined process pond for eventual recov­ery of the ammonia in the fertilizer granulation plant. The heavy fog lines used more water than was needed to control vapors and generated water· faster than could be pumped to the process pond. There was also concern about the capacity of the plant ponds to hold all the generated water.

To solve the problems, the fog nozzles were replaced by lawn sprinklers. Canvas tarps and burlap bags were laid on the ground around the tank to help trap and re­strict the movement of the ammonia vapors. On top of the tarps were placed soaker hoses, long flat lawn hoses with perforations along its entire length. The soaker hoses placed strategically around the tank were successful in removing most vapors and generated much less water than the fog nozzles.

Immediately after detection of the leak, an around­the-clock operation began to transfer the contents of the tank into trucks for transportation off site. The ammonia was transported to atmospheric storage tanks at the J. R. Simplot facility in· Helm, California and at the California Ammonia Company facility in the Port of Stockton. The entire transfer operation took 3 days, from 6:00 A.M. on Tuesday to 10:00 P .M. on Thursday.

Actions were taken to prepare for a potential cata­strophic failure of the tank. In this worst-case scenario, the dike area around the tank would fill up with liquid ammonia which would then begin to vaporize and drift into downwind areas. To control the rate of vaporization from a pool of liquid ammonia, ways were investigated to insulate the pool from the air and the sun. Two drums of F413 foam (HAZMAT FOAM No. 1) were ordered for application onto the pool surface. However, the foam would cover an area of 1,200 sq. m. (13,000 sq. ft.) for only an hour. To find an· alternative to the foam, poly­urethane was tested. The plant lab floated a piece of polyurethane sheet on top of some liquid ammonia to test for a reaction and discovered that the only problem was some absorption of ammonia into the sheet. Therefore, the plant sent an emergency order for four hundred 1.2 m (4 ft.) by 2.4 m (8 ft.) by 2.54 cm (1 inch) thick polyurethane foam sheets. In case the tank had failed, the polyurethane sheets would have been placed on top of the ammonia pool to reduce the amount of exposed surface area and to insulate the ammonia until it could be properly removed.

Emergency Response

In California, state law requires that emergency response agencies be notified immediately following a release or potential release that presents an immediate threat to the health and safety of employees or the public. Following the initial evaluation, the Simplot Lathrop facility emergency coordinator notified the National Response Center in Washington, D.C., the Lathrop Fire District, and the County Air Pollution Control District. These notifications were made at 6:30 A.M. The fire district responded immediately to the scene, evaluated the situation, then made notifications to the county Office of Emergency Services (OES), the Sheriff’s office, and the Manteca City Fire Department. Further notifications were made by the county OES to other groups and agencies.

Public perceptions

Public perceptions of the incident were shaped by statements from company and public officials through the local media. The media received information from the J. R. Simplot Company, the city of Manteca, and the OES emergency operations center. Public information releases were not coordinated between these different sources. Four newspapers covered the incident in detail: the Stockton Record, the Modesto Bee, the Manteca Bulletin, and the Manteca News. The incident also received local televi­sion coverage. Stockton radio station KJOY 1280 and Manteca cable television station 10 were used as public emergency information stations.

Overall, the newspaper articles covering the incident were accurate and objective; however, the articles were often topped with sensational headlines.

Comments and reactions from the general public varied from the uninterested to the paranoid. After the standby evacuation notices were distributed, some people called the Lathrop Fire Department wanting to know if it was safe to return home from work or from a local shopping trip. A few people called the Lathrop Water District because they thought their groundwater was affected by the leak. Another resident was afraid the tank would explode. Others wanted to leave town. One person said, ”I was scared to come back to work in the afternoon because I was afraid to leave the pets.” Another said, “We have our grandson staying with us, so maybe I should just leave and go to my daughter’s house for the night.” One resident said that the incident “cut too close to home.”

Other members of the community were not as concerned.

The ammonia tank leak incident covered a span of three days. During those three days, no one was hurt, killed, or hospitalized. No violation of local air pollution regulations occurred, no nuisance fumes were released out­side of the plant, and no degradation of the environment occurred. From a technical standpoint, the ammonia tank leak was merely a nuisance leak, a minor event. But from a public view, it was a major event which commanded the attention of more than a dozen agencies, dominated the news headlines, and created much concern within the surrounding community.

The lack of a unified command among the various jurisdictions covering the incident contributed to the increased concern within the surrounding communities.

Another factor that increased public concern was the lack of a combined public information source. The two main sources of public information were the emergency operations center run by the OBS and the public informa­tion officer from the city of Manteca. Communication between the two information centers was limited, result­ing in some misunderstandings. For example, each command post drafted its own copy of the standby evac­uation notice which were worded differently from each other. Some residents received both notices, creating unnecessary confusion.


In order to evaluate how the emergency was handled by the company, four general questions were asked.

Could the incident have been prevented?

Because of errors during fabrication, tank failure was destined to happen and probably could not have been avoided. Although containment dikes were constructed around the tank to contain a major spill, a small secon­dary containment wall constructed close to the tank wall, like a tank within an open tank design, could have significantly eliminated the vapor release from the leak and made it easier to control. To avoid future leak incidents, periodic inspections of the floor welds could be done, but this procedure would put restrictions on tank operation. Another option would be to store ammonia in smaller quantities at the facility. This would reduce operational flexibility and requires daily movement of ammonia; however, it would minimize the impact of a tank failure on the local community.

How well was the chemical release controlled?

The methods used in managing the ammonia fumes were successful in keeping the fumes within or close to the facility fence line. The ammonia vapors were captured by water spray from fog lines or lawn sprinklers. The water was contained within a dike system, where it was pumped to a process pond for eventual recovery of the ammonia in the fertilizer granulation plant.

How well did the company communicate with emergency officials?

The facility has an excellent working relationship with the Lathrop Fire District. The fire department knew how to respond to an ammonia incident and acted appropri­ately. The company had a poor relationship with the city of Manteca. The Manteca fire department had little or no experience with ammonia, and there was no line of com­munication between city officials and company managers.

Could the company have minimized the negative impact on its image?

A review of the leaking ammonia tank incident has given some insight into the management of chemical re­lease incidents. A facility should be prepared for any type of chemical release on its property, and should have per­sonnel ready to respond and to handle all human and tech­nical emergencies. It should also have a plan to work with public agencies and to deal with the surrounding com­munities. In this way, a company can exercise some con­trol over an emergency situation.

To help minimize the impact of any future incidents, the following steps may be taken.

First, know your neighbors. The facility should have a good working relationship with all local agencies and jurisdictions, and most important of all, a strong line of communication must exist between them. During an emergency situation, the facility and the various agencies should feel comfortable contacting each other. This will help to minimize the impact of an incident by allowing for the flow of good information and by preventing misunderstandings.

Second, provide good information and provide it quickly. Immediately after an incident begins, the company should supply as much information as possible to the agencies and to the public. If the information needed by the agencies and desired by the media is not provided, they may gravitate towards other sources outside of the company. It is better for information to come from qualified sources than to let bystanders speculate. How­ever, the information and recommendations given should be relevant and well thought out and should not be misleading nor speculative. Bad information or the lack of information may create distrust and cause overreaction or bad judgement.

Third, the company must present itself as a team. Immediately after an incident begins, the company should be prepared to give strong, solid-recommendations on how to handle the incident, what precautions need to be taken, and how much of any area will be affected. This infor­mation should be presented to ·the agencies and media by a facility team so that any question presented can be properly and competently addressed. The team ought to include people who can answer technical questions deal­ing with plant operations and maintenance, chemistry, the environment, and public health and safety. The team also should have people who can answer questions dealing with legal issues and public relations.

After the incident

After the cause of the leak was found, the tank was taken out of service. The decision was made to store all anhydrous ammonia in four fixed pressurized bullet tanks with a total capacity of277 metric tons. The atmospheric tank has been converted into a parts storage building, affectionately referred to by employees as the “igloo” (figure 6).


Greiner, M. L., “Emergency Response Procedures for Anhydrous Ammonia Vapor Releases,” AICHE Plant/Operations Progress (1984).

Husa, H. W. and Bulkley, W. L., “Hazard of Liquid Ammonia Spills from Low Pressure Storage Tanks,” American Oil Company, Whiting, Indiana.

  1. R. Simplot Company, “Lathrop Plant File 0467 -Ammonia Tank Leak Failure” (1986).

Manteca Bulletin (1986).

Modesto Bee (1986).

Stockton Record (1986).

San Joaquin County Office of Emergency Services, “Haz­ardous Substance Emergency Report” 1986.

Testing Engineers, Inc., ”Examination and Evaluation of Cracks in Ammonia Tank Floor Plate,” Project No. 14284 (1986).


We believe that sharing knowledge by constant publishing of relevant articles can enhance the ammonia Process Safety Management (ammonia PSM) awareness and contribute to ammonia and urea industry employees working together to identify the root causes of accidents, minimizing risks and prevent incidents from repeating