Hazards of Oversizing a Rupture Disk 4

[Pavan Kumar]

Dear Friends,

I would like to understand the hazards in over sizing a Rupture disk. I have to size a Rupture disk for a few scenarios. The most conservative case is that of an internal explosion which results in 12" size for the rupture disk. For failure of regulator case the ruptures disk required is 8". The reason for the 12" size for the internal explosion is more because of the higher relieving temperature ( 3500 Deg C) as opposed to 580 Deg C for Failure of regulator scenario.

If I select the 12" size then the Rupture disk is over sized for the Failure of regulator scenario which is more feasible than the Internal explosion scenario. If select 8" size then it is undersized for the Internal explosion scenario which though is quite unlikely.

I would like to understand the hazards if any of using the 12" size as it covers for both regulator failure scenario and the internal explosion scenario. The set pressure is 34 psig.

Thanks and Regards,
Pavan Kumar

 

[RVAmeche]

Presumably both the 8" and 12" disc options would be set at the same set pressure? The difference is one will protect your vessel for all scenarios and the other will not.

If you expect to experience frequent regulator failures or process upsets, a separate relief valve for those scenarios may be prudent since it re-closes.

I can't think of any issues with the 12" disc, aside from increased cost of the disc and piping. But that's what your determinant case requires, so it is what is. If you have a regulator failure with a 12" disc, the velocities through the 12" piping will be correspondingly lower.

 

[Pavan Kumar]

Hi RVAmeche,

Yes the set pressure is same for the both scenarios at 34 psig. My worry is that whether the 12" rupture disk will burst at 34 psig for the regulator failure scenario because the RD sheet material is designed thick for higher temperature and the energy required to burst may not be enough in the case of regulator failure scenario. I shall also consult the Rupture disk manufacturers regarding this problem.

Thanks and Regards,
Pavan Kumar

 

[EdStainless]

I have never done the design work but I have seen installations with two rupture discs with the relief pressures slightly different from each other. The thought was that a minor event would only blow the lower set one and a major event would blow both.
Though RVA'a suggestion of looking at a relief valve plus rupture disc makes a lot of sense.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy

 

[RVAmeche]

When talking to the rupture disc manufacturer be sure you include the different temperatures. The difference will affect the burst point, but hopefully it won't be too extreme. I'm sure they have methods of making it work for both scenarios.

 

[Pavan Kumar]

Hi RVAmeche,

I will definitely mention to the manufacturer about the different relieving temperatures for the applicable scenarios and their impact on the RD size. I will post the learnings on this thread.

Thanks and Regards,

Pavan Kumar

 

[don1980]

Pavan, i think you need to re-evaluate your temperature values. They look way too high, and if you use these values your disks will effectively be "set" for a much higher burst pressure. If/when an explosion happens, the disk is going to burst due to the pressure wave created by the explosion, and this occurs in a fraction of a second, long before the disk is heated to the temperature of the gas that is generated by the explosion. For this scenario i think you should specify a burst temperature that corresponds to the temperature in the vessel before the explosion occurs. 3500 C is an extraordinarily high temperature. I doubt that any vendor will even consider providing a disk at a burst temperature this high.

You haven't said anything about the process that you're protecting, but I'm also skeptical of the temperature you cited (580 C) for the regulator failure case. I don't know what type of process you're dealing with so i can't say that this is wrong, but that's an exceptionally high temperature for a regulator failure case. is this process really operating at this temperature?

One of the problems with using a rupture disk rather than a PRV is that disks open based on pressure and temperature, whereas PRVs just based on pressure (temperature has a small effect but it's not significant). When you have an explosion scenario you have to use a disk - PRVs don't open fast enough.

Anytime you use a disk in an application that has multiple scenarios with widely differing relief temperatures, then you have a problem. If you spec the burst temperature based on the scenario with the lower temperature, then that disk will burst at a lower pressure if/when a high temperature scenario occurs. if you spec the disk burst temperature based on the high-temperature scenario, then the disk will burst at a higher pressure if/when the low temperature scenario occurs. There's no way to solve this problem - you can't get a disk that will reliably burst at a fixed pressure over a wide range of temperatures. That's just one of the problems you have to be willing to accept when you choose to use a rupture disk, and it's one of the reasons to avoid the use of disks.

 

[TiCl4]

I agree with Don here - 3500 C is above the melting point of all metals that I know - even tungsten clocks in at 3,422 C. I assume you calculated the combustion temp of an internal deflagration and applied that temp to the disk. There will not be enough time to heat the disk significantly before the pressure wave pops it - it should be set for the normal operating temperature of the vessel when the deflagration occurs.

Also - very important here - are you sure you mean "explosion"? Deflagrations occur when the flame propagation speed is less than the speed of sound in the medium. Explosions/detonations occur when the flame speed exceeds the speed of sound in the medium. In the case of detonation, the pressure rise is very localized. Detonations have the possibility of over-pressuring the tank walls near the explosion source before the shock wave even touches the rupture disc. In this case, providing a rupture disc would not protect your vessel from rupture. If you truly have an explosion, not deflagration, potential, providing a rupture disc may not provide any protection.

 

[Pavan Kumar]

Hi don1980 and TiCl4,

I fully agree with your responses. I too feel that the rupture disk burst pressure is dependent on the temperature. I spoke to Zook disk company and the sales person told me that the maximum temperature their rupture disks can tolerate is 2700 Deg C and he is not sure if he can set the burst pressure to that we are asking. We will hear more from him soon.

I apologize for not describing the process earlier. Let me describe it here now. The Process we are doing is preheating Process Air to 350 Deg C using a Natural gas burner. The Preheated air enters our reactor for reactor-start-up operation. The Natural gas is supplied through a 1.5" line, the flow of which is controlled by a regulator. We are providing the rupture disk on the Process air line to protect the reactor process side from over pressure.

The over pressure scenarios that I have identified are for more than normal flow of Natural gas entering the Burner and getting ignited. This can be due to failure of the Natural gas regulator whose opening gets adjusted based o the downstream pressure. I have added the increased heat due to combustion of Natural gas to the Process air and calculated the new temperature at 592 Deg C. The relief rate is the Process air flow( m1 lb/hr) plus the Natural gas flow ( 321 lb/hr) to be relieved at 592 Deg C. The Natural gas flow of 321 lb/hr is estimated based on the pressure differential of 8 psi across the regulator with this its wide open CV. I used the Kr method and calculated the RD size at 10" for this scenario.

Second and the controlling scenario is based on our past internal explosion(deflagration) incident.
In this scenario I calculated the Natural gas flow through the 1.5" pipe for the known pressure differential of 8 psi without the regulator. I estimated this to be at 3260 lb/hr. Now combusting this and adding the heat to the air results in a temperature of 2700 Deg C. So the relief rate and relief temperature are now at (m1+3260) lb/hr at 2700 Deg C. This scenario may not be possible because there is a regulator in place that will restrict the flow due to its CV. I did this calculation because of the past internal explosion that happened earlier and I want to accommodate the RD for this case.

The RD size calculated for this scenario is 12".

The question that arises is how will I get a rupture disk that bursts at 34 psig for a relief temperature of 2700 Deg C and still bursts at 34 psig for a relief temperature of 592 Deg C.

Both scenarios are to protect against overpressure due to internal explosion or deflagration. This will last for just a few seconds. The pipe temperature is not expected to rise that quickly. You may be right the pressure wave bursts the rupture disk before the temperature can rise to either 592 Deg C or 2700 Deg C.

When I use 350 Deg C as relief temperature an 8" RD is found it be adequate for both the scenarios. The question then arises is am I specifying the correct temperature for the selecting the disk material.

I would be keen to hear your opinions.

Thanks and Regards,

Pavan Kumar

 

[Latexman]

No regulator is far fetched, isn't it? I mean, where does it go? How does it get taken out of service and a connection made from one side to the other?

Use the wide open Cv.

Second, it seems to me your best indicator of problems is temperature. Develop a high integrity safety system to shutdown the NG supply when temperature deviates too much from normal. You can get the same level of protection as a RD with some redundant temperature measurements and final control elements. It seems you want to use something similar to the burner management system in a fired process.

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[TiCl4]

"because of the past internal explosion that happened earlier " !!!

What happened that resulted in this "internal explosion"? How did this happen, and what was the result (pressure/temperature rise experienced)?

Also, something is not clear to me. Is this a direct contact heating? I.e. is the process air in direct contact with the flame, and the process air and combustion products all go to the reactor? Is that how you are getting your relief rate?

"So the relief rate and relief temperature are now at (m1+3260) lb/hr at 2700 Deg C"

Something doesn't add up. If you have enough oxygen to completely burn all of the incoming natural gas to achieve that temperature, then there is nothing left to cause a deflagration. You would instead just have an overtemp/overpressure possibility (due to increased volumetric flow rate) that would need to be relieved. However, there is nothing in this case that would go "Boom", and you stated there was an internal explosion earlier.

If instead your case natural gas pushes the mixture above the UEL, which then later ignites, your relief rate is NOT dictated by your steady-state flows. Instead, it is a dynamic event (a deflagration pressure wave) that can't be sized simply by looking at your starting flow rates.

Maybe I'm missing something and someone can illuminate me, but I don't quite understand the initiating scenario the OP presented.

 

[Latexman]

Unless it is direct contact and they lack a good burner management-like control system. They flamed out and kept feeding? No flame detectors? It didn't light and they kept feeding? No purge timers? PK, please tell us the root cause of the "explosion". And, was it a deflagration (not an explosion) or a detonation (explosion)? A deflagration can overpressure a vessel and it may rupture (it may sound like an explosion).

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[Pavan Kumar]

Hi TiCl4,

I am not sure how the past internal explosion happened. All I know is that it happened. I am not able to get more details on it. All my problems are because I want to accommodate it into the possible scenarios.

The Process Air is heated by direct flame due to Natural gas ignition. The Process air and combustion products go to the reactor. This is not important as this is just preheating step and the Process air to NG mass flow rate ratio is around 245:1. The sketch is located here

https://files.engineering.com/getfi...574b&file=Preheater_RD_Schematic_Sketch-1.pdf

Because the air to fuel ratio is so high the combustion is complete. You are right there is no Natural gas left for deflagration. But how does one explain or calculate the relief rate for the internal explosion incident that happened. The pressure rise may be due to temperature rise on heating air in a fixed volume. But the there is no obstruction to the flow as the pipe is open to the compressor on the upstream and to the reactor on the downstream side which is goes to scrubbers and then to a atmospheric stack. How can pressure build up in this case, unless it is a deflagration. For the 321 lb/hr possible due to failure open or wide open CV of the Natural gas regulator the air fuel molar ratio is less than 5%, the LEL. The relief temperature however is 592 Deg C. This however will be 7% when I can flow 3260 lb/hr Natural gas through the 1.5" NG pipe without the NG regulator. You may question how this is possible, but then I want to simulate the case that caused the internal explosion earlier. The relief temperature now becomes 2700 Deg C.

In short I am having difficult developing the rationale for the scenario and if I try to accommodate something that may not be possible, I run into high relief temperature issue.

I will look for your opinion to resolve this issue.

Thanks and Regards,

Pavan Kumar

 

[Latexman]

Exactly where did the explosion occur? Where was the damage? Upstream of preheater, downstream of preheater, in RX D, downstream of RX D?

What else goes to RX D?

It reminds me of a plant I used to work at making cyanide with the Andrussow process.

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[Compositepro]

You may be getting confused/bogged-down by irrelevant details. In a deflagration in a closed volume the worst case pressure scenario is with a stoichiometric mixture of fuel and oxidizer. The peak temperature pressure go down as you move away from a stoichiometric ratio. The energy released in a deflagration is usually limited by the amount of oxygen present. I'm not sure how you are using flow rates to get any useful conclusion. A deflagration is over in less than a second.

I've done the calculation before and stoichiometric mixtures will increase in pressure by a factor of about ten during a deflagration. With a detonation the shockwave will have a higher peak pressure, but a rupture disc will do nothing to mitigate a shockwave.

 

[TiCl4]

I think you may want to reconsider your initiating scenario. It seems much more likely that a burner failure or gas leak caused an abnormally high gas concentration, which subsequently ignited. You yourself have acknowledged that the proposed scenario would not result in deflagration. I would instead explore scenarios of fuel/air mixtures igniting in the preheater.

Also, how do you keep the flame going? You say a failed-open regulator will still have the gas concentration below the LEL of 5%. This implies normal natural gas flow is even less and would be well under the LEL. How is the flame sustained?

Internal deflagrations in preheaters are problematic to handle. They are usually prevented by a high-integrity Burner Management System, and upstream/downstream consequences are controlled via blowout panels or explosion isolations, to name a couple of options.

 

[don1980]

S**t happens, especially when: (1) humans are involved, and (2) they're lighting burners. That's why burner management systems are essential for safety. If you don't have one already, you need to get one.

Based on the additional details provided, the only potential scenario I see is deflagration. You're showing 18 psig upstream of the regulator, so a failed open regulator by itself isn't going to overpressure the system. If that fuel ignites, then of course that's a different story, but that's a deflagration and not "regulator failure".

Protection from deflagration is explained in NFPA 68. Depending on the size of the system, that may be done with a traditional rupture disk, or by use of an explosion panel which is effective a very large rupture disk. Regardless of whether you use a rupture disk or an explosion panel, this device needs to be installed where the ignition occurs (on the burner housing) and not on the upstream air line where it's shown on your sketch. With deflagration you're operating on a millisecond time scale - you don't have time for the pressure wave to travel upstream to reach the disk.

Whether you use a rupture disk or an explosion panel, these devices are going to opened by the pressure wave while the device is still at it's normal temperature.

 

[Pavan Kumar]

Hi Latexman,

The explosion occurred on the Rupture disk which was upstream of the Preheater. The Rupture disk along with its holding flanges flew into air and landed some 100 ft away. Luckily it fell when no one was around, else there would have been injuries or even fatalities.

Thanks and Regards,
Pavan Kumar


Latexman (Chemical)
21 May 20 22:03
Exactly where did the explosion occur? Where was the damage? Upstream of preheater, downstream of preheater, in RX D, downstream of RX D?

What else goes to RX D?

It reminds me of a plant I used to work at making cyanide with the Andrussow process.

 

[Pavan Kumar]

Hi Compositepro,

The molar ratio of Oxgyen to Methane is greater than 2:1, so the stoichiometric ratio required for deflagration is present. In fact we have excess Oxygen. The heat released due to the combustion raises the temperature of the air. I am calculating the relief rate and relief temperature for two different scenarios.

1. Regulator Failure : In this scenario the Natural gas(NG) regulator is assumed to fail open and the flow rate calculated for wide open CV. This amounts to 321 lb/hr. The mol% of NG in air is less than 5%, the LEL for Methane. I am still assuming that deflagration will occur. The heat released due to the complete combustion 21526 Btu/lb is added to the Process air. This raises its temperature from 350 Deg C to 592 Deg C. Process air plus releasded NG flow rate equal to ( m1 + 321 ) lb/hr is to be relieved from the Rupture disk at 592 DEg C. This is resulting in disk size of 10".

2. Other - Past Incident Learning : In this scenario I am calculating the mass flow rate of Natural gas that will flow in 1.5" pipe for the normal pressure differential of 8 psi by assuming that all the fittings are in place except that the NG regulator is not in place. This may sound ridiculous. The reason for doing this is to get the temperature of the Process air around 2000 Deg C as I feel the temperature would have risen to this value in the Past internal explosion event. I do not know what caused the internal explosion. All I know is that it happened. I wanted to get the Rupture disk to be sized for this situation also.

The flow rate through the 1.5" pipe is calculated as 3260 lb/hr and the Process air gets heated to 2700 Deg C. The relief rate and relief temperature are therefore ( m1+3260 ) lb/hr and 2700 Deg C. This is resulting in a disk size of 12".


From what I understand from discussion with my esteemed colleagues in this thread that the burst pressure of the Rupture disk is dependent on the temperature. So it is difficult have 34 psig, my desired burst pressure for the rupture disk for both these scenarios as the relief temperatures are vastly different. I also understand that since the deflagration is over within a second, the Process air will not get a chance to heat before the pressure wave pops open the disk, so the relief temperature to be considered is 350 Deg C, the normal preheating air temperature. When checked the RD size for both the above scenarios is 8".

I did not understand this statement "I've done the calculation before and stoichiometric mixtures will increase in pressure by a factor of about ten during a deflagration.". Could you please elaborate.
We are considering deflagration or internal explosion and not detonation which required explosion panels instead of rupture disk. I guess we are ok there. You said you did these kind of calculations earlier. Could you please share your learning on sizing rupture disks for the scenarios I elaborated above.

I now need to conclude so that I can write my scenario rationale and the size the rupture disk that caters to the internal explosion that happened earlier.

Thanks and Regards,
Pavan Kumar


Compositepro (Chemical)
21 May 20 22:39
You may be getting confused/bogged-down by irrelevant details. In a deflagration in a closed volume the worst case pressure scenario is with a stoichiometric mixture of fuel and oxidizer. The peak temperature pressure go down as you move away from a stoichiometric ratio. The energy released in a deflagration is usually limited by the amount of oxygen present. I'm not sure how you are using flow rates to get any useful conclusion. A deflagration is over in less than a second.

I've done the calculation before and stoichiometric mixtures will increase in pressure by a factor of about ten during a deflagration. With a detonation the shockwave will have a higher peak pressure, but a rupture disc will do nothing to mitigate a shockwave.

 

[Pavan Kumar]

Hi TiCl4,

The burner failure and excess NG leaking is equivalent to my high NG flow that results in 7 mol% of NG in the Process air which when ignited leads to deflagration. The question is now about the relief temperature. Should I use the 2700 Deg C which results due to the heat combustion adding to the Process air or should I sue 350 Deg C, the normal preheating temperature as the deflagration raises the pressure too quickly for the temperature to rise.

For the normal case the NG mol% in air is around 0.7 %. I think for ignition the mol% doesnot need to be at LEL or above LEL. It will ignite but not explode I guess. If the concentration is between LEL and UEL then it is in explosive zone. Please correct me if I am wrong.

Internal deflagrations in preheaters are problematic to handle. They are usually prevented by a high-integrity Burner Management System, and upstream/downstream consequences are controlled via blowout panels or explosion isolations, to name a couple of options.

Rupture disks per API 521 can handle internal deflagrations but not detonations which require explosion panel sized as per NFPA 68. Please correct me if I am wrong here.

The excerpt from API 521 is copied below.

If overpressure protection is to be provided against internal explosions caused by ignition of vapor-air mixtures where the flame speed is subsonic (i.e. deflagration but not detonation), rupture disks or explosion vent panels, not relief valves, should be used. These devices respond in milliseconds. In contrast, relief valves react too slowly to protect the vessel against the extremely rapid pressure buildup caused by internal flame propagation. The vent area required

is a function of a number of factors including the following:
a) initial conditions (pressure, temperature, composition),
b) flame propagation properties of the specific vapors or gases,
c) volume of the vessel,
d) pressure at which the vent device activates,
e) maximum pressure that can be tolerated during a vented explosion incident.

It should also be noted that the peak pressure reached during a vented explosion is usually higher, sometimes much higher, than the pressure at which the vent device activates.

Design of explosion-relief systems should follow recognized guidelines such as those contained in NFPA 68 [118].Simplified rules-of-thumb should not be used as these can lead to inadequate designs.

If the operating conditions of the vessel to be protected are outside the range over which the design procedure applies, explosion vent designs should be based on specific test data, or an alternate means of explosion protection should be used.

Some alternate means of explosion protection are described in NFPA 69 [119], including explosion containment,explosion suppression, oxidant-concentration reduction, and so forth.

Explosion-relief systems, explosion containment, and explosion suppression should not be used for cases where
detonation is considered a credible risk. In such cases, the explosion hazard should be mitigated by preventing the formation of mixtures that could detonate.Explosion prevention measures, such as inert gas purging, in conjunction with suitable administrative controls can be considered in place of explosion-relief systems for equipment in which internal explosions are possible only as a result of air contamination during start-up or shutdown activities.


Thanks and Regards,

Pavan Kumar



TiCl4 (Chemical)
21 May 20 22:55
I think you may want to reconsider your initiating scenario. It seems much more likely that a burner failure or gas leak caused an abnormally high gas concentration, which subsequently ignited. You yourself have acknowledged that the proposed scenario would not result in deflagration. I would instead explore scenarios of fuel/air mixtures igniting in the preheater.

Also, how do you keep the flame going? You say a failed-open regulator will still have the gas concentration below the LEL of 5%. This implies normal natural gas flow is even less and would be well under the LEL. How is the flame sustained?

Internal deflagrations in preheaters are problematic to handle. They are usually prevented by a high-integrity Burner Management System, and upstream/downstream consequences are controlled via blowout panels or explosion isolations, to name a couple of options.