October 01, 2005
Cascading Failures
The accident that happened yesterday in my laboratory did not have any serious consequence - except some economic damage - but it's a good exaple of how a failure in a complex system can trigger a series of events ending in serious trouble. The accident wasn't fault of anyone in particular either; laymen always want to find a culprit but engineers know better.
However, before explaining the accident in detail, I have to introduce a few basic concepts of reaction engineering. For a chemical reactor, conversion is the fraction of reactants that is converted into products, and it depends from the residence time (among other factors): residence time is defined as the average time the reactants spend inside the reactor; it is easily calculated dividing the reactor volume by the volumetric flow of the reagents. In my case, the reactor volume is 300 mL, and with a reactant flow of 300 mL/min the residence time would be 1 minute.
Generally, research reactors work at low (10% or less) conversion. There are a number of reasons for this; one is that at high conversion it is not possible to study certain phenomena. But the main reason is that in case of exothermic reactions, keeping the conversion low will produce only a little amount of heat, amount that can easily be handled without recurring to cooling systems. To accomplish this, the simplest way is to have short residence times, and thus pretty high feed flow - compared to the reactor volume.
Another thing to notice is that the catalysts used in industry tend to deactivate - they lose their catalytic activity with time. This can be due to contamination with substances that bond to the active sites on the catalyst (sulphur is the most common), or to modifications of the physical/crystal structure of the catalyst itself. Obviously, chemical industries are interested in finding ways to reactivate their catalysts in situ, because it potentially saves money.
Thus, this Big Chemical Industry contacted my supervisor and his job was to find a way to reactivate a catalyst being used by this BCI. For the task, he used a Berty reactor, that is a sort of sturdy metal container (made at least of stainless steel, but probably of something even more resistant to chemical attacks) with a bolted cover, a stainless steel or copper seal, a heating system and a stirrer. Yes, because if the contents of a reactor are well mixed, one does not have to worry about concentration inhomogeneties inside the reactor itself. For this reason, the catalyst is placed in a fine mesh basket, in turn fitted inside an impeller that can turn at different speeds thanks to an electronically controlled electric motor.
To avoid the severe complication of a high-temperature, high-pressure seal around a rotating shaft, the impeller is driven magnetically: the shaft connected to the motor rotates a magnet just underneath the bottom of the reactor, and this magnetic field in turn rotates the impeller (that probably has a magnet embedded into it too). However, even with this design the stirrer shaft gets hot, and it rotates on a water-cooled bearing. Total cost of the new rig, around £40 000.
The furnace heating system has three zones each one with an indipendent controller, plus a fourth controller for the interior of the reactor itself, connected to a thermocouple placed inside the reactor. All these controllers work together to keep the temperatures at the set levels within less than 1 K deviations - and are set to shut off the heaters if the temperature exceeds a certain setpoint.
So, my supervisor performed a certain procedure for the reactivation of the catalyst and when it was completed he begun feeding reactants (a mixture of hydrogen and carbon monoxide, I think - I'm not working on that project) to check whether the catalyst was reactivated properly.
But at certain point, for reasons under investigation, the automatic shutoff valve on the feed line tripped and shut off the reactant flow - or better, allowed only a very small floe through it. With a very small flow, the residence time increased sharply, and the reaction went to almost 100% conversion, producing a large amount of heat. The temperature in the reactor increased to 300 C in an occurrence known as runaway reaction, and even if the automatic protections shut off the heaters it was too late: heath transferred through the stirrer shaft damaged its bearing. My supervisor could do nothing but turning all the whole thing off, and feeding the reactor with a mixture of hydrogen and nytrogen while it cools down. Now, he's stuck with a non-working reactor loaded with a pyrophoric catalyst and so it cannot be opened to check for eventual damage, and it is not possible to run reactions in order to proceed with the project - and the BCI isn't very happy about that. The catalyst had been reactivated, because an inactive catalyst does not cause a runaway reaction, but there are no useful data to measure the catalyst activity. The stirrer shaft bearing will have to be replaced, but the shaft itself may have been scratched. And the rig was produced by an US company that has no service centres in Europe.
So, a stupid incovenience caused cascading failures that led to a rather serious accident - that fortunately did occur during the day and not at night. And if things are already bad enough on this scale, on industrial scale a runaway reaction can cause serious economic damage, and even catastrophic explosions with several casualties. For these reasons, chemical plants have features such as relief valves, emergency cooling systems, nitrogen injection lines (nitrogen is inert, so it will flush the reactants away and help with cooling) and even reaction inhibitors injection systems.
Update 03/20: No, the catalyst basket does not rotate; only the impeller at the bottom of the reactor does. But the catalyst basket has vertical baffles at its periphery to generate turbulence and hence improve mixing. However, the graphite bearing of the impeller disintegrated in the accident, and it needs to be replaced - with considerable loss of time and expense.
There is another aspect to runaway reactions: the rate of chemical reactions increases with temperature, so when a reaction produces excessive heat (you can see it as power expressed in Watts) it also becomes faster, thus producing even more heat in a vicious circle.
However, before explaining the accident in detail, I have to introduce a few basic concepts of reaction engineering. For a chemical reactor, conversion is the fraction of reactants that is converted into products, and it depends from the residence time (among other factors): residence time is defined as the average time the reactants spend inside the reactor; it is easily calculated dividing the reactor volume by the volumetric flow of the reagents. In my case, the reactor volume is 300 mL, and with a reactant flow of 300 mL/min the residence time would be 1 minute.
Generally, research reactors work at low (10% or less) conversion. There are a number of reasons for this; one is that at high conversion it is not possible to study certain phenomena. But the main reason is that in case of exothermic reactions, keeping the conversion low will produce only a little amount of heat, amount that can easily be handled without recurring to cooling systems. To accomplish this, the simplest way is to have short residence times, and thus pretty high feed flow - compared to the reactor volume.
Another thing to notice is that the catalysts used in industry tend to deactivate - they lose their catalytic activity with time. This can be due to contamination with substances that bond to the active sites on the catalyst (sulphur is the most common), or to modifications of the physical/crystal structure of the catalyst itself. Obviously, chemical industries are interested in finding ways to reactivate their catalysts in situ, because it potentially saves money.
Thus, this Big Chemical Industry contacted my supervisor and his job was to find a way to reactivate a catalyst being used by this BCI. For the task, he used a Berty reactor, that is a sort of sturdy metal container (made at least of stainless steel, but probably of something even more resistant to chemical attacks) with a bolted cover, a stainless steel or copper seal, a heating system and a stirrer. Yes, because if the contents of a reactor are well mixed, one does not have to worry about concentration inhomogeneties inside the reactor itself. For this reason, the catalyst is placed in a fine mesh basket, in turn fitted inside an impeller that can turn at different speeds thanks to an electronically controlled electric motor.
To avoid the severe complication of a high-temperature, high-pressure seal around a rotating shaft, the impeller is driven magnetically: the shaft connected to the motor rotates a magnet just underneath the bottom of the reactor, and this magnetic field in turn rotates the impeller (that probably has a magnet embedded into it too). However, even with this design the stirrer shaft gets hot, and it rotates on a water-cooled bearing. Total cost of the new rig, around £40 000.
The furnace heating system has three zones each one with an indipendent controller, plus a fourth controller for the interior of the reactor itself, connected to a thermocouple placed inside the reactor. All these controllers work together to keep the temperatures at the set levels within less than 1 K deviations - and are set to shut off the heaters if the temperature exceeds a certain setpoint.
So, my supervisor performed a certain procedure for the reactivation of the catalyst and when it was completed he begun feeding reactants (a mixture of hydrogen and carbon monoxide, I think - I'm not working on that project) to check whether the catalyst was reactivated properly.
But at certain point, for reasons under investigation, the automatic shutoff valve on the feed line tripped and shut off the reactant flow - or better, allowed only a very small floe through it. With a very small flow, the residence time increased sharply, and the reaction went to almost 100% conversion, producing a large amount of heat. The temperature in the reactor increased to 300 C in an occurrence known as runaway reaction, and even if the automatic protections shut off the heaters it was too late: heath transferred through the stirrer shaft damaged its bearing. My supervisor could do nothing but turning all the whole thing off, and feeding the reactor with a mixture of hydrogen and nytrogen while it cools down. Now, he's stuck with a non-working reactor loaded with a pyrophoric catalyst and so it cannot be opened to check for eventual damage, and it is not possible to run reactions in order to proceed with the project - and the BCI isn't very happy about that. The catalyst had been reactivated, because an inactive catalyst does not cause a runaway reaction, but there are no useful data to measure the catalyst activity. The stirrer shaft bearing will have to be replaced, but the shaft itself may have been scratched. And the rig was produced by an US company that has no service centres in Europe.
So, a stupid incovenience caused cascading failures that led to a rather serious accident - that fortunately did occur during the day and not at night. And if things are already bad enough on this scale, on industrial scale a runaway reaction can cause serious economic damage, and even catastrophic explosions with several casualties. For these reasons, chemical plants have features such as relief valves, emergency cooling systems, nitrogen injection lines (nitrogen is inert, so it will flush the reactants away and help with cooling) and even reaction inhibitors injection systems.
Update 03/20: No, the catalyst basket does not rotate; only the impeller at the bottom of the reactor does. But the catalyst basket has vertical baffles at its periphery to generate turbulence and hence improve mixing. However, the graphite bearing of the impeller disintegrated in the accident, and it needs to be replaced - with considerable loss of time and expense.
There is another aspect to runaway reactions: the rate of chemical reactions increases with temperature, so when a reaction produces excessive heat (you can see it as power expressed in Watts) it also becomes faster, thus producing even more heat in a vicious circle.
Comments:
Post a Comment
