Dates: Apr 1, 2021
ioMosaic presents a case study on a vapor leak involving a polymerization runaway reaction in May 2020 at an industrial facility in India, sadly leading to several deaths and several hundred injuries. We developed a dynamic model of the incident, which evaluated the effectiveness of monomer inhibitors. We used our model to deconstruct the potential causes that we then compared with actual incident reports. Our article concludes that mixing can lead to hot thermally stratified layers and hence poor cooling in monomer storage tanks, where inhibitor is not mixed effectively to suppress the polymerization reaction. These factors were all found to have contributed to the incident.
The following is an abridged version of an article published in The Chemical Engineer magazine April 2021 edition.
A vapor leak involving a polymerization runaway reaction occurred in May 2020 at the M/s LG Polymers Pvt Ltd in R. R. Venkatapurm, Vizakhapatnam district, India. The plant had been closed as part of a national lockdown to prevent the spread of COVID-19. One of the styrene monomer tanks, whilst inhibited, was involved in a runaway reaction spread of COVID-19. One of the styrene monomer tanks, whilst inhibited, was involved in a runaway reaction with several fatalities and many hundreds of hospitalisations.
The suspected runaway reaction scenario was modelled using Process Safety Office® SuperChems™. Using experimental data, depletion models can be developed for the inhibition of the styrene and applied to ioMosaic’s qualified styrene thermal polymerization kinetic model. This method can This method can be applied to different kinetic models and scaled-up to simulate real life events.
To model the effects of solar radiation and heat transfer, the tank was divided into segments. This allowed us to accurately model the impact of hot days and the heat loss occurring during the night. Note that the top of the tank and walls show significant change in temperature because they are not in contact with the liquid contents.
Figure 1: Wall heat transfer dynamic model predicted duration till runaway
Our model divides the tank up into segments so we can geometrically model heat transfer from the surroundings to the vessel contents. The initial concentration of inhibitor (15ppm) and temperature (17°C).
From our modelling we concluded that the reaction most likely occurred towards the top of the tank within a thermally stratified layer separate from that measured by the temperature probe at the bottom of the tank, which was still measuring 17°C moments before the time of the runaway reaction occurring due to insufficient circulation of cooled styrene.
We therefore assumed that effective refrigeration unit was lost, or the tank was thermally stratified, on Day 40. This modelled scenario confirms that lower amounts of inhibitor was mixed within these layers and was not well mixed throughout the tank. Our model predicts that for an inhibitor concentration of between 3 and 7 ppm the polymer concentration was consistent with actual measurements as shown in the figure below. This agrees with the conclusions reached in published independent reports on increases in polymer concentration measured at the top of the tank.
Figure 2: Predicted inhibitor concentration in stratified layers
Our simulations confirmed that several factors could have contributed to the incident: (1) Piping modification, therefore preventing adequate circulation in the tank, (2) thermal stratification of the tank which includes stratification of inhibitor and poor inhibitor management, (3) procedural failures in monitoring of tank polymer concentrations. Our modelling also confirmed that the tank temperature must be consistently kept low for the inhibitor induction time to remain high (so for TBC, a predicted maximum temperature of 20°C) and inhibitor concentrations must remain at acceptable levels, well-mixed throughout the tank to prevent stratification.
The Chemical Engineer Magazine published the full article, Deconstructing Runaway Reactions Using Dynamic Modelling, by James Close, John Barker and Georges A. Melhem, Ph.D., FAIChE, in their April 2021 edition. Read the full article now.