Runaway Polymerisation of Glacial Acrylic Acid
The Incident
A 1.5 te batch of glacial Acrylic Acid was charged to a feed vessel from drums. To this was added ca.200 kg of n-Dodecyl Mercaptan (nDM). The vessel contents should have been immediately transferred to a reactor containing a heel of water and methanol but, due to delays in obtaining raw materials, this transfer did not take place. 4.5 hours after charging the nDM a violent runaway reaction occurred. The bursting disc on the feed vessel ruptured and polymer was sprayed out of the works and into a nearby, on-site car park. Several cars were damaged and one person was taken to hospital suffering from the effects of vapour inhalation.
The Investigation
Carius tube screening tests revealed that nDM lowered the onset temperature of thermal polymerisation of acrylic acid by ca.70 K.
The acrylic acid used at the time of the incident was adequately inhibited with 210 ppm MEHQ. However, the normal plant procedure for charging glacial acrylic acid is by placing the feed vessel under vacuum and sucking the monomer in. It was shown experimentally that de-aeration of the acrylic acid (which may arise as a result of the charging method) destabilised the monomer - this is unsurprising since MEHQ requires the presence of dissolved oxygen to remain effective.
In adiabatic Dewar studies the feed vessel contents at the time of the incident showed accelerating exothermic behaviour from ambient temperature. After 1.5 hours the temperature had increased to ca. 80°C, at which point the temperature increased rapidly to 160°C in under 5 minutes. Polymeric material was violently ejected from the Dewar flask.
Conclusions
The incident arose as a result of glacial acrylic acid being stored in the presence of a highly reductive mercaptan for extended periods of time. The monomer had been further destabilised as a result of the charging process.
Lessons to be Learnt:
- Glacial acrylic acid should be pumped from drums to plant vessels, NOT charged by vacuum or pressure.
- If vacuum charging is to continue as a Company practice then provision must be made to ensure that the monomer is re-aerated in the feed vessel.
- Ultimately the process must use a separate feed vessel for nDM.
- Monomer feed vessels should be fitted with temperature indicators. At the time of the incident there was no way of knowing that the temperature of the monomer feed vessel contents was rising.
- Acrylic acid and nDM must not be charged to the feed vessel before all raw materials have been obtained.
- Acrylic acid and nDM must only be charged to the feed vessel when immediate transfer to the reactor can be guaranteed.
- The practice of using pressure vessels as monomer feed vessels is again being questioned. We were fortunate in that we managed to (successfully?) relieve this runaway. A different monomer or different rate of polymerisation could have resulted in the feed vessel becoming over-pressurised or even rupturing.
- Process operator training on the causes of potential hazards is needed.