Microsoft Word - 025.docx


 CHEMICAL ENGINEERING TRANSACTIONS  
 

VOL. 48, 2016 

A publication of 

 
The Italian Association 

of Chemical Engineering 
Online at www.aidic.it/cet 

Guest Editors: Eddy de Rademaeker, Peter Schmelzer
Copyright © 2016, AIDIC Servizi S.r.l., 
ISBN 978-88-95608-39-6; ISSN 2283-9216 

Drum Burst due to Runaway Reaction and the Limits of MOC 
Alexis Pey 
Swissi Process Safety GmbH, Mattenstrasse 24, CH-4002 Basel 
alexis.pey@tuev-sued.ch 
 
Daily operations and equipment failures may require slight process changes which in many cases are carried 
out without any effect on safety. But sometimes unknown hazards are hidden beyond those slight changes.  
Even if a company has a well implemented Management of Change protocol, the evaluation of every single 
process change may be difficult to assess in detail as a balance between time available and analysis depth 
has to be achieved, therefore assessment is made with information and knowledge available in the spot. In 
this sense, even if the evaluation is done, the knowledge available may lead to neglect hazards.  
In this paper, an example of a slight process change is presented which lead to a runaway reaction and a 
drum burst. Hopefully no injures had to be blamed and only damages on the area where the drum burst had to 
be considered.  
A solvent and a process reactant were mixed for a too long time inside a drum due to a pump failure leading to 
a different operation during the dosing of reactants into the reactor.  
The change was evaluated according the MOC and available criteria by plant operator, shift leader and 
production chemist on duty, without identifying any significant hazard thus allowing the chance to be carried 
on.  
Investigations later revealed that a hazardous exotherm reaction was present at temperatures close to 
ambient temperature between the reactant and the solvent.  
This case is an example of an unknown by the operational workforce – known only by experts and the fact that 
experts may not be systematically involved in the assessment of slight process changes due to the time 
available to carry out operations.  
Beside the analysis and explanation of the accident, the aim is to define measure to prevent similar incidents 
to take place; therefore lessons learned allow defining some recommendations related to MOC management 
which can be applied to any company helping to unhide hazards and make the knowledge available to people 
carrying out operations in the field.  

1. Description of the operations 

At this point, the operation as designed is presented, then the process modification and its reasons are 
explained and, finally, the incident and its consequences are described. 

1.1 Intended operation 

During a chemical synthesis process, Sulphuryl Chloride is used as reactant in an intermediate step. 
This substance is received in drums, which are handled in groups of 4 on a pallet. The normal production size 
requires 4 drums to be dosed into the reactor. 
Dosing of Sulphuryl Chloride is done by using a dip pipe introduced in the drum and pump sucking out the 
substance and pumping it to the reactor. 
The pumping rate is limited in order to control the exotherm released during the reaction and keep a low 
reactant accumulation. In this sense, the dosing time for each drum was set to 2 hours, meaning that the 
dosing operation of Sulphuryl Chloride lasts around 8 hours per batch.  
When no more Sulphuryl Chloride could be pumped out from the drum, and in order to rinse the few litters 
remaining inside, 20L of Xylene were introduced in the drum without stopping the dosing pump to the reactor. 

                               
 
 

 

 
   

                                                  
DOI: 10.3303/CET1648135

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Please cite this article as: Pey A., 2016, Drum burst due to runaway reaction and the limits of moc, Chemical Engineering Transactions, 48, 
805-810  DOI:10.3303/CET1648135  

805



Operational details at this step are important. As the drums had two orifices on the top, the bigger orifice was 
used to introduce the dip pipe sucking out the Sulphuryl Chloride while the smaller one was used to pump in 
the 20L of rinsing Xylene. 
Once the 20L of Xylene had been introduced and no more liquid could be pumped out, the drum was 
considered as completely used and therefore closed and handled as a dirty drum for disposal. 

1.2 Process deviation 

When the dosing of Sulphuryl Chloride started on the day of the incident, it was noticed that the pump used to 
transfer the Xylene for rinsing was not running well. As the reparation could not be done immediately, the 
situation was assessed and the process was modified as described next. 
With only one pump in operation, it was decided that once no more Sulphuryl Chloride could be sucked out 
from a drum, the transfer would be stopped and then dosing pump would be used to introduce the 20L of 
Xylene into the almost empty drum, after which it would be immediately closed. The drums with the Sulphuryl 
Chloride and Xylene mixture would remain closed until the last drum was charged; at this moment the pump 
woud be used to transfer to the reactor the rinsing mixture in all 4 drums. 
When the process modification was evaluated, it was thought that is was safer to keep the Sulphuryl Chloride 
diluted with Xylene in an almost empty drum than just leaving a few litres of pure reactant inside. A reason 
considered at that moment was that Xylene, besides being used for rinsing, was as well the main solvent in 
the reaction mass and, therefore, it was thought that no reactive hazards would be present in the mixture 
resulting from diluting a few litres of Sulphuryl Chloride with the process solvent. 

1.3 Incident 

Proceeding as described above, the Sulphuryl Chloride in three drums was completely transferred and all 
three were filled with 20L of Xylene and closed. 
While the fourth drum was being dosed to the reactor, meaning that the other 3 drums were on the same 
pallet, closed and only containing the leftover of Sulphuryl Chloride and rinsing Xylene, the operators noticed 
that the second drum (in order of dosing) was becoming pressurized. A clear deformation could be observed 
on that drum. 
Initially, operators attempted to carefully loosen the cap and release the pressure, but as the attempts were 
not successful they decided to stop the transfer on the fourth drum and evacuate the area. 
A few minutes after the zone was evacuated, the drum exploded. 
The other two empty drums were blown apart, remaining only in the pallet the last drum, which was full, and 
fortunately did not suffer any damage causing a leak.  
The incident resulted in no personal injuries. Neither fire nor major spill was reported. Damages were limited 
to the pallet and neighbouring drums as well as some equipment in the loading area, such as piping for local 
vapour exhaust and transfer pipes. 

2. Reactivity of Sulphuryl Chloride 

According to the description of the incident, it was obvious that the reactivity of Sulphuryl Chloride had 
probably played a clear role on the event. In this sense, and as already mentioned above, the fact that the 
mixture between the reactant and the solvent could show a dangerous behavior was unknown at that time by 
the company. 

2.1 Amount of Sulphuryl Chloride 

In order to assess the reactive behaviour, the mixture composition is initially estimated.   
In first place the amount of Sulphuryl Chloride remaining inside the drum once it is considered as empty is 
estimated. To do so, next considerations were made: 

A. Drum was tilted when empty, but this operation depends on the skills of the operator, the tilting 
degree and the time during which the drum is hold in a tilted position. According to experience it is 
considered reasonable to assume that a thin layer of 0.5 cm of liquid remains inside the drum when it 
is declared empty. The amount of substance that this layer represents in a 55 cm diameter drum with 
a chemical having a density of 1.67 (g/cm3) at 20ºC is 1,19L or 1983g. 

B. The configuration of the piping caused that once the dosing pump was stopped a small amount of 
substance was drained back to the drum. The estimation made by considering the layout of the 
system lead to a result of around 3 litres of liquid able to flow back to the drum. A volume of 3 L in a 
substance with a density of 1.67 (g/cm3) at 20ºC represents a mass of 5010 g. 

Therefore, a total amount of around 7kg of Sulphuryl Chloride (4.2L) could remain in the drum when the 20L of 
Xylene were introduced for rinsing. 

806



Considering a density of Xylene of 880 (kg/m3) at 20ºC the mass of Xylene added with rinsing purposes is 
17.6 kg. 
As a result a mixture with a concentration of 28.5% w/w of Sulphuryl Chloride in Xylene was obtained after 
introducing the rinsing solvent in each empty drum. 

2.2 Reactivity with water 

Nevertheless, it is well known that Sulphuryl Chloride reacts violently with water releasing gas, therefore the 
presence of water was investigated in relation with the incident. 
At this point the operations and experience acquired while handling this substance are of importance. 
According to operational experience, when the relative humidity is high, Sulphuryl Chloride drums rapidly 
generate fumes when opened. This behaviour is caused due of the reactivity of the substance with moisture 
contained in the air. 
Being aware of this behaviour, on the day of the incident it was decided that the drums would be closed 
immediately after charging the rinsing Xylene. According to the explanations of operators, when this operation 
was performed no fumes were observed and no water was present at the loading station. 
Moreover, as the reaction with water is very fast and violent, in case that water could have entered the drum, 
the reactivity would have been observed much earlier. The explosion occurred around 4h after charging the 
drum contents into the reactor, introducing the Xylene for rinsing and closing it. 
Finally, it is reasonable to consider that the drum was tightly closed since when the pressurization started it 
was not possible to open it. 

2.3 Reactivity with organics 

As can be found in literature, although Sulphuryl Chloride is supplied in galvanised drums they may be unsafe 
and lead to unintended reactivity with this substance.  
It is known that Sulphuryl Chloride can chlorinate many active organics, including aromatics, ketones and 
probably esters. This will release sulphur dioxide and hydrogen chloride, a fairly soluble gas, which will react 
with zinc and iron to produce hydrogen, an insoluble gas, and zinc or iron chlorides, both catalysts for many 
reactions. 
As mentioned before, the reactive potential of Sulphuryl Chloride often involves gas generation, moreover 
documentation available describing the reactivity of this substance with organics and galvanised drums, 
triggers the suspicion that it is not a safe process condition to store Sulphuryl Chloride with Xylene in a 
galvanised drum for an extended period of time (a few hours). 

2.4 Experimental tests 

Finally, thermal stability tests were done by means of DSC in order to investigate and confirm the suspicions 
on the potential exothermal behaviour of a mixture between Sulphuryl Chloride and Xylene. 
To do so, mixtures of Sulphuryl Chloride and Xylene at different concentrations were investigated in both 
stainless steel and gold crucibles. 
As a summary next table shows the most relevant results obtained for a mixture with a concentration of 25% 
w/w of Sulphuryl Chloride in Xylene: 

Table 1: DSC 25% w/w Sulphuryl Chloride in Xylene. Gold Crucible 

Variable Value Unit Comments 
Crucible Material Gold --  
Left Peak Temperature 147.01 ºC  
Peak Temperature 190.68 ºC  
Right Peak Temperature 205.50 ºC  
Heat Released 102.59 J/g Normalised value 

Table 2: DSC 25% w/w Sulphuryl Chloride in Xylene. Stainless Steel Crucible 

Variable Value Unit Comments 
Crucible Material Stainless Steel --  
Left Peak Temperature 107.09 ºC  
Peak Temperature 176.37 ºC  
Right Peak Temperature 195.40 ºC  
Heat Released 139.77 J/g Normalised value 
 

807



3. Pressure build up 

As a rule of thumb, it is considered in practice that the bursting pressure of a standard 220L metallic drum is 7 
to 10 bar. 
Reactions related to Sulphuryl Chloride may easily release gas; nevertheless in a DSC test the pressure 
evolution inside the cell is not available. 
In first place, as the mixture showed to have a certain exothermal behaviour, it is evaluated if the adiabatic 
temperature rise could lead to the drum pressurisation due to the vapour pressure exclusively due to the 
substances present in the mixture. 

3.1 Vapour pressure 

Considering the composition of the mixture in the drum and the liquid heat capacity of the substances,  the 
resulting mixture heat capacity is around 1.6 kJ/(kg·ºC), which, linked to a heat release around 140 J/g and 
considering adiabatic conditions would lead to a maximum temperature increase of around 88ºC. 
Considering that the operation was being done at ambient temperature, the final temperature inside the drum 
would be around 105ºC. 
In figures 1 and 2 the vapour pressure curves as a function of temperature are respectively shown for 
Sulphuryl Chloride and o-Xylene. As can be seen, at a temperature around 105ºC, and considering the typical 
bursting values considered for a 220L drum, the vapour pressure alone would have not been able to cause the 
incident with the consequences observed. 

 

Figure 1: Vapour pressure curve of Sulphuryl Chloride 

3.2 Reaction gases 

In all potentially considered unintended reactions of Sulphuryl Chloride, the gases generated are hazardous 
for the health and the environment, therefore the operation of opening a pressurized drum with this substance 
inside it is considered as very dangerous. 
As well, as the pressure rise rate and the current pressure inside the drum are unknown during this type of 
events, it should be considered that the drum may burst at any time, and therefore, staying close to it or trying 
to do some actions on the drum is not considered as a safe behaviour. 

808



3.3 Bursting a Drum 

In order to reach a pressure of 10bar inside a 220 L drum, considering ideal gas behaviour and 22.4 L per 
mol, around 100 moles of gas are required. 
In this sense, when Sulphuryl Chloride decomposes, it may generate 2 moles of gas per mol of Sulphuryl 
Chloride. Considering that Sulphuryl Chloride has a molecular weight of 134.97 (g/mol), the amount of 
Sulphuryl Chloride required to generate 100 moles of gas is approx. 6750 g or 4 L. 
This calculation does not aim to conclude that the reaction involved in the accident was the decomposition of 
Sulphuryl Chloride, but only to show that the amount of Sulphuryl Chloride required to pressurise an empty 
drum is not very large and could be achieved easily with only a few litres of substance. 

 

Figure 2: Vapour pressure curve of o-Xylene 

4. Management of Change 

In terms of Management of Change (MOC) the protocol was applied as defined in the safety management 
system, but the problem laid on the knowledge available at the time of evaluating the situation. 
The assumption that a reactant is stable when mixed with a solvent lead to the wrong conclusion that the 
mixture between Xylene and Sulphuryl Chloride would not represent a reactive hazard if kept at room 
temperature for an extended process time. 
Lessons in this sense were introduced indicating that in case of a longer process time, the stability of the 
chemical mixture should be carefully evaluated and a more in depth information review should be performed in 
front of a similar situation. 
The instability of Sulphuryl Chloride when handled in metal drums as well as its reactivity with aromatic 
substances could be found in literature, but these sources were not checked. In this sense a more detailed 
specification of required resources to be considered while investigating reactive hazards was also defined. 

5. Conclusions 

To start with conclusions related with this incident, it should be said in first place that  the evaluation of a 
process deviation requires always an in depth and detailed review, otherwise, process hazards may remain 
unidentified leading to dangerous process conditions. 
In this sense MOC protocols are very important and they should also be as detailed as possible, showing 
required information sources, people to involve and hazards to be evaluated. 

809



The fact that a hazard is known will not prevent that we continue learning from accidents as long as it remains 
unknown for us. In this sense, prevent dealing with know/unknowns should a clear target in any safety 
management system. 
In relation with the incident itself and its consequences, as Sulphuryl Chloride is used in many chemical 
processes involving different types sectors, the information contained in this report and the incident that took 
place may be useful for any company handling or using this substance in its processes. 
Finally, during the accident investigation it was also shown how important is to have a minimum testing 
capacity in site. Screen equipment such as a DSC brings a solid basis on the assessment of thermal stability 
of chemical mixtures. In this case the DSC was made to late, as it was made to investigate an incident, which 
brings us back to realise how important is to have good MOC procedures.  
 

References 

Bretherick's Handbook of Reactive Chemical Hazards, 6th edition, 1999. 
CCPS, 1995, Guidelines for Chemical Reactivity Evaluation and Application to Process Design, AIChE. 
CCPS, 1999, Process Safety in Batch Reaction Systems, AIChE. 
CHETAH v.9.0, ASTM 
OEM -U.S. EPA, ERD-NOAA & CCPS, Chemical Reactivity Worksheet v.2.0.2 
Stoessel, F., 2008, Thermal Safety of Chemical Processes: Risk Assessment and Process Design. WILEY-

VCH. 
SuperChems Software v.6.40mp, ioMosaic 
 

810