(104l) Explosion In An HMX (Octogen) Plant

1.      DATE AND TIME OF INCIDENT (When?):  

01/June/09 7:00 am.

2.      INCIDENT LOCATION (Where?):

a.  Company name:    Austin Powder International

b.  Company facility:

HMX Plant – Rafaela – Santa Fe – Argentina.

c. Process outline:

This modified Bachmann process for the manufacturing of HMX has been operated for almost 20 years without incident.

There are several stages in the batch process to produce the final recrystallized product The incident occurred in the nitration/reaction vessel. The vessel is a simple water jacketed vertically mounted reactor with a vertical stirrer.  This is a batch process involving the reaction of hexamine and nitric acid/ammonium nitrate in the presence of acetic acid and acetic anhydride. Once all the ingredients have been added to the nitrator the reaction continues for some time – the temperature is controlled to approximately 44 ⁰C using the water jacket. Once the reaction is complete the vessel is discharged by gravity through a series of hand operated valves to another vessel.  The lines are purged by compressed air after each discharge operation. The chemistry of the process is complex and several by-products are also produced in very small quantities -  these by-products are not well characterized. This stage of the process results in the preparation of the crude HMX as a suspension. There are a few other operations in different parts of the world following the same or a similar reaction principle and incidents have been very rare. HMX in suspension is considered to have a low sensitivity to friction and impact.

To increase the throughput (initial design required discharge of the reactor every 2 hours), the process was modified in 2007 by installation of a ‘post reactor’ to allow the reaction of the crude mixture to complete (after 1 hour) in this reactor and to start a new batch in the main reactor every other hour.

3.       DESCRIPTION OF INCIDENT (What happened?):

On the day of the incident 8 batches of HMX had already been processed. There was nothing unusual reported by the plant personnel. The operator had just finished discharging batch 9 to the next vessel and was in the process of clearing the lines and valves when the explosion occurred in the discharge lines/valves. The precise actions being undertaken at the moment of the explosion are not clear but from statements made and the positions of valves it is thought that the explosion coincided with or immediately after the opening of the valve on the compressed air line.

The explosion did not propagate beyond the valves and is thought to have involved a very small quantity of material.

After attending to the injured operator and inspecting the plant, the remaining crude HMX was processed in the normal fashion.

4.       IMPACT OF INCIDENT (What effect?):

One operator was injured and sent to the hospital, another was slightly injured.

The discharge pipe and 3 valves were damaged. The entire material damage was small. There was no damage to buildings or structures.

There were no other health or environmental effects.

5.       LIKELY CAUSE OF INCIDENT (Why did it happen?):

Operating procedures were followed at all times.

The most likely causes were either impact or friction on the by-products that form during the process.

Impact may have been caused by small metal particles from inside piping accelerated by air that was used to clear blocking in the pipe to transfer the suspension for separation/filtering.

An ignition of accumulated by-product on the inside surfaces of the valves/connecting pipe by friction is also possible when the operator turned a ball valve.

The risks of more sensitive intermediates in the mixture transferred from the main to the post reactor and accumulation of these in piping and valves and compressed air causing friction or impact (in accelerating small solid particles (from internal corrosion)) was not considered.

Based on the experience with the almost two decades running failure-free single reactor system, there was confidence that process control was safe.

6.       ACTIONS TO PREVENT RECURRENCE (How to prevent it?):

As the chemistry in this complicated process is not fully understood, the process flow was studied in detail.

The reactor outlet was enlarged and metal piping was replaced with a reinforced plastic hose (less confinement and to mitigate the consequences in the event of reaction/explosion).

The compressed air line has been completely removed.

The total number of valves has been reduced and all ball valves were replaced with diaphragm valves that will involve less potential for friction on any deposited product.

The routine maintenance procedures have been improved to reduce the potential for deposition of product in lines and valves.

The whole process was reviewed applying a HAZOP study. This led to the implementation of further safety devices to deal with other hazards that did not contribute to this incident.

The inventory of samples was reduced. These are now kept in a separate location that cannot be hit by shrapnel (it was not the case, but could have happened).

7.     LEARNING POINTS (What can we learn from this incident?):

Wet materials (HMX and by-products) were not as insensitive as it was assumed.

Processes operating for a longer period of time should be reviewed regularly in applying Hazard Studies (and HAZOP where applicable).

Basis of Safety documentation should be reviewed regularly vs. new knowledge available from literature or other resources.