Is there any special procedure for PSV load calculation in fire case when relifing pressure above Critical pressure?

You may refer to "Rigorously Size Relief Valves for Supercritical Fluids" by Ryan Ouderkirk.

You may also get some clues from this link..."Determine Latent Heat for Multi-Component and Relieving Area Using Rigorous Method in HYSYS"

Dear Joe

Thanks for response.

I studied the mentioned article has issued by Ryan Ouderkirk. But I have some question about it

1. Is it necessary to consider our vessel full of liquid or HLL shall be considered for calculation of wetted-area?
2. For first point I should to consider operating temperature at relieving pressure of bubble pint temperature at relieving pressure?

But second reference (chemical & process technology information) let me to study it , then I will discuss with you.

Thanks very much

Mohammad,

Above critical point, the fluid has gone into dense phase. If you are interested see how a fluid change from liquid to vapor and finally gone into dense phase, check out "How does Supercritical fluid looks like ?".

Entire vessel will be occupied by dense phase fluid. At dense phase, the fluid possibility presents liquid and vapor like properties. Thus, conservative approach is to consider entire surface area is wetted.


If you read "Determine Latent Heat for Multi-Component and Relieving Area Using Rigorous Method in HYSYS", you may bring the temperature from operating condition to relieving condition (supercritical in your case).

Joe:

I have worked with CO2 for a lot of years and have produced and handled all 3 phases – gaseous, liquid, and supercritical fluid (SCF). I am very interested in what is now being presented as a problem: the behavior of the SCF in a fire scenario with subsequent pressure relief through a PSV.

We never had the luxury of having thermochemical properties for fluids in my younger years, so all we could do was speculate about how the SCF would behave through a PSV. Recent information indicates that the SCF behaves very much as a gaseous fluid and therein lies my concern for what we are discussing. My concern is in the form of a question: does the film heat transfer coefficient for the SCF behave as one for a gaseous fluid? And if so, can we expect that the heat transfer from an external fire would have to struggle with an extremely weak and inefficient heat transfer mechanism across such a film?

I think that you know where I'm going with this: if the heat transfer from the external fire is indeed going to find a difficult path to find a heat sink within the vessel containing a SCF, then we've got a more serious problem – that of a potential vessel wall stress failure and eventual rupture. This then, would be a direct and logical case for a vessel depressurizing and not one for a PSV to save the vessel.

I have seen the effects of CO2 cylinders and even fire extinguishers exposed to external fires and fortunately for us all this has never been a vessel rupture case because by the Grace of God (I can't credit anyone else) all CO2 cylinders and fire extinguishers have been equipped with a rupture disc since even before I was born – which was a little after the Jurassic Era. So the Old Timers never had this problem – we solved it with one of Phil Leckner's prime relief tools: the rupture disc.

Because of the above uncertainties, I would question if we should (in the case of a SCF fire scenario) reconsider if the "conservative" design approach is really one of considering that the entire vessel surface area is wetted internally and that we would experience efficient heat transfer and a subsequent proportional pressure increase. I don't think liquid boiling or transformation to the vapor state is a consideration here.

My design approach when dealing with the above scenario in the past has been one of applying a rupture disc or a depressurization. I wouldn't put any trust on predicted SCF properties unless convinced otherwise, and the rupture disc philosophy in my opinion is the most conservative solution.

Although I consider this a rare application for many, it is interesting to exchange points of logical process design in how to attack this type of problem. I await anyone's valued comments.

Mr Montemayor,

I hope i understood your concerns.

Relief & Blowdown load
The entire discussion here was much related to methodology to find the relief load. In my point of view, once the fluid get into supercritical region, it will behave like vapor and/or liquid. Although some information indicating that it is more behave like gaseous, nevertheless so far i have not read any article proved to me that it is a gas. In the present of this uncertainties, i would rather consider a conservative approach where the supercritical behave like "liquid" and entire vessel is wetted with "liquid". Those heat input into the vessel is at maximum. This would results a conservative relief load.

In the same way, in the event a Blowdown system is present to protect the system, the fire blowdown rate would be conservative as well.

Above consideration mainly to provide a conservative approach to derive the relief load and blowdown load.

Internal pressure induced stress vs vessel allowable stress
On the hand.........when calculating the vessel stress level, i would consider the supercritical is behave close to "gaseous". Same opinion as yours, gaseous in the vessel having very low heat transfer. It just like a "insulation" layer inside the vessel. Heat input from external fire would possibly stay in the vessel metal, increase the temperature and reduce vessel stress level. Very minimum heat will be transferred into the fluid and majority (or all) of the heat will stay and increase the vessel temperature (weaken the vessel).

The blowdown rate will be designed such that any point at a time, the stress caused by internal pressure will always lower than the vessel stress. This is the way to safeguard the vessel from catastrophic failure.

You may noticed that above consideration is a mix of two non-aligned approach. This mainly due to lacking of proper understanding and accurate properties prediction of supercritical fluid (i am referring to myself). If any of you are familiar and can accurately predict the behavior of supercritical fluid, a new and less conservative approach may be used.


Rupture Disc (RD) vs PSV
If the internal pressure possibly reach the RD or PSV set pressure, then probably a RD is better choice.

However, in my point of view, a vessel containing gas or "gaseous" like fluid i.e. supercritical fluid, the vessel is potentially fail before the internal pressure is reached Rupture disc (RD) or PSV set pressure. I would think other protective measures like blowdown system, external cooling, etc may helps to minimise the likelihood of catastrophic failure. In many event, i would emphasize on blowdown system.

One of the downside of RD which i always concern about is high probability of failure due to erosion and fatigue failure and lead to inventory loss, subsequent shutdown of the system, production loss, long restart...which is very costly.

For the CO2 cylinder and fire extinguisher case, i would properly agreed with RD solution as this CO2 is utility and supporting the production (i guess). There will be a back-up unit to avoid total plant shutdown and blowdown. Loss CO2 has limited cost. However, if the plant is producing CO2, i will reconsider to use RD.

I am quite agree with your approach on blowdown. I would emphasize more on blowdown system, external cooling (deluge), fire extinguishing facilities, etc rather than put attention on the relief load.

Dear Joe

Thanks for your kindly attention.....i applied your procedure ..i will send it for you ..please check it..& let me your openion.

thanks

Joe:

Thank you for the prompt reply. Your efforts (& I hope mine also) will help Mohammad attack his problem with a better insight and approach in order to develop a conservative solution to a very complex and arcane problem.

My intent here is to emphasize the fact that we can't discuss a Latent Heat of Evaporation principle on an application that is well outside of the saturated "dome" found in the Mollier or T-S Diagrams. When a fluid is in the super critical phase, there simply is no "liquid" phase to "evaporate". There is only one, continuous, "dense" phase that apparently behaves sometimes like a gas. Therefore, I have to challenge the idea of using any equation that employs the concept of latent heat or liquids. Knowing this, I would apply caution and skepticism to the idea that we can use the prior API 521 equations like a recipe.

I recommend caution because even the API, in its latest 5^th Edition & addendum of May 2008, makes too much use of the qualifying words "might" and "may" when prescribing a new method of approach for relieving "vessels containing only gases, vapors or super critical fluids" in Section 5.15.2.2.2. If you and others don't have this edition & addendum yet, I highly recommend it be obtained. This whole subject is one that has come out of the dark engineering closet to haunt us with the realization of our ignorance and lack of understanding on this application and serves the purpose of concerning us.

Reading through the above has made me happy in my ignorance because it finally comes to a point where we, as engineers, have come to admit as much regarding those very complex and un-researched areas of our expertise. We simply don't know EVERYTHING! We are human, and we are now somehow admitting it. The equations offered for this application (Equations 8 through 12) are based on very loose principles and are simply "estimates" (to use the exact API term) at best. I won't detail them out – nor the logic – because I would want everyone reading this to thoroughly read the new API 521 edition.

I am very proud of the "old timers" who taught me and mentored me so many years ago. They, indeed, recognized the very basic complexity of this application and devised a pure, conservative, fail-safe method of dealing with it and proceeding onwards to other pressing needs. They applied a rupture devise that works to depressurize a pressure vessel containing a super critical fluid and mitigated a disaster. To get a "feel" of how sure and confident the API is about their presentation of equations, read their Section 5.15.2.3, "More rigorous calculations":

"If the user considers that the preceding assumptions in 5.15.2.2 are not appropriate, more rigorous methods of calculations may be specified. In such cases, it can be necessary to obtain the required physical properties of the containing fluid from actual data or estimated from equations of state. It might be necessary to consider the effects of vessel mass and insulation. The pressure-relieving rate is based on an unsteady state. As the fire continues, the vessel-wall temperature and the contained-gas temperature and pressure increase with time. The pressure-relief valve opens at the set pressure. With the loss of fluid on relief, the temperatures further increases at the relief pressure. If the fire is of sufficient duration, the temperature increases until vessel rupture occurs.

Procedures are available for estimating the changes in average vessel-wall and contained-fluid temperatures that occur with time and the maximum relieving rate at the set pressure [51], [52]. These procedures require successive iteration. For fire-insulated segments exposed to fire, it is recommended to assume the fire temperature outside the vessel wall. The heat-transfer resistance from the wall to the fluid is very low compared to the insulation layer's resistance and can be (is usually) neglected. A more rigorous method is described in Reference [141]."

By the way, there are 16 pages of Bibliography in this API edition of 200 pages.

Mr. Montemayor,

I am sure the efforts (always) that you have put in will definitely benefit the reader of this post. I just hope my contribution would trigger people to think more...

I am completely understand and agree with your intent to emphasize the fact that we can't discuss a Latent Heat of Evaporation principle in this context. The supporting statement is "When a fluid is in the super critical phase, there simply is no "liquid" phase to "evaporate". There is only one, continuous, "dense" phase that apparently behaves sometimes like a gas.".

Somehow supercritical fluid may not 100% behave like gas with low conductivity, low specific heat and low heat transfer. Let say in "Measurements of Heat Transfer Coefficients From Supercritical Carbon Dioxide Flowing in Horizontal Mini/Micro Channels", and "Heat Transfer and Pressure Drop of a Supercritical Pressure Fluid Flowing in a Tube of Small Diameter", they presented the Cp has singificant increase for certain pressure and temperature range. Experiment results by Yamagata et al. (1972) in the "CFD Analyses of Heat Transfer to Supercritical Water Flowing Vertically Upward in a Tube" presented the heat transfer coefficient for supercritical fluid is high in certain temperature and pressure range. All these uncertainties have lead us to a question, should we ignore these known characteristic ?

Equation (8) to (12) are rather weak in the sense that it leads us to a condition no way you can know how much heat is entering the vessel and what kind of heat flux you will expect. Backward estimate would help us to understand the heat flux is low. Is this approach conservative ? Before i aware the special characteristic as indicated above, i would believe supercritical is gas like. But when i aware of these special characteristics for supercritical fluid, i just has NO clue with correct approach.

The only reasonable way that i can tackle is to take conservative heat input. The magic equation by API (21000A^0.82) is known to provide more conservative heat flux / heat input. This has also been recommended in the paper "Rigorously Size Relief Valves for Supercritical Fluids" presented by Ryan Ouderkirk.

One of the point that we are still struggling is the magic equation by API is base on Pool fire. Pool fire is known to have much lower heat flux than jet fire. Again, the question surfaced. Is the API magic equation conservative enough ? When we consider the jet fire impingement, the metal temperature would increase rapidly. In some/many events, the vessel fail prior to the internal pressure reach it RD or PSV set pressure. This has lead to INVALID case.

Well... I am not as lucky as you where "old timers" taught you and mentored you for many years ago. I am shame to say this but it is a fact. We only can read more, think more, take many time and effort to understand a simple concept and phenomenon....I am glad that you have the luxury of "old timers" mentoring and guiding you. On the other hand, i am glad that there are still some people like you willing to spent your time and take effort to provide guidance to us...

API use a lot of "may" and "might". Statement like "If the user considers that the preceding assumptions in 5.15.2.2 are not appropriate,..." will only lead us to think extra and find a new way to tackle the problem.

Looks forward more advice and taught from you.

Thanks in advance.

Much of this conversation has centered around whether a SCF is more "gas like" or "liquid like". I don't believe much is to be gained by focusing on the difference between gas and liquid. Obviously, by coincidence, the gas phase for one substance may have (many) physical properties similar to the liquid phase of a totally different substance. Diferences in specific physical properties is normal and expected, so we should not use values of physical properties to distinguish between gas-like and liquid-like. Instead, the chief difference as it pertains to relief valve sizing is that liquids would typically vaporize when large amounts of heat are introduced; gases, of course, do not exhibit this phenomenum. Vaporizing liquids exhibit high rates of heat transfer and are an excellent heat sink since latent heats are typically much higher than specific heats over moderate temperature ranges. These, I would contend, are the salient factors to be considered when approaching this problem. Looking at it in this way, we must discard the equations we normally use for sizing liquid containing vessels exposed to fire, since we have no phase change. Thus SCF calcs are "single phase like".

BTW, if you dislike rupture disks for any reason, you may want to consider an alternate technology such as a rupture pin valve for relief.

Doug,
Thanks for your opinions and participation.