Process Wednesday: "Why Reactions Run Away"

Apropos of nothing, I'm doing a little reading on runaway reactions (and how to prevent them from happening.) Here's some joyful little anecdotes from the process chemistry literature [1]:

Case History 1. An explosion occurred in a nitration reactor. Although the company had carried out small-scale development tests, they had not measured the heat generated by the reaction, and the cooling system was inadequate. A particular problem associated with scale-up is the assumption that the so-called “onset temperature” for exothermic runaway or thermal decomposition will be the same in a small-scale test as in a full-scale plant. It is not always appreciated that the temperature at which such runaways/decompositions occur is dependent on a number of factors including the detection sensitivity of the test apparatus used,
vessel size and heat transfer characteristics, and time.  

Case History 2. An explosion occurred in a process vessel involving a thermally unstable material. The company had made the assumption that the decomposition temperature of the material was in the range 270-299 °C, the same as in small-scale tests reported in the literature. Following the incident the company carried out a fuller investigation of the thermal decomposition characteristics of the material in more accurate adiabatic tests, designed to simulate the plant environment. These tests indicated that the material could decompose at 153 °C on the plant scale, below the temperature of the heating jacket.

The lecture transcript from the OPRD article by J.C. Etchells suggests that thermal runaways have 3 root causes:

• inadequate understanding of the reaction chemistry or kinetics
• underrating of the safety controls or backup systems 
• inadequate procedures or training

One notes that the T2 explosion happens to cover 2 of these quite easily (poor understanding of the reaction, and failed backups on the cooling.) All pretty fascinating, and pretty terrifying. 

1. Etchells, J.C. "Why Reactions Run Away." Org. Process Res. Dev. 1997, 1, 435-437.