Dynamic simulation vs steady state simulation and PSV sizing

[rocketscientist]

We're having a lively discussion in our department where and when to apply a dynamic simulation for sizing a relieve valve. I've argued that dynamic is not necessary for a fire scenario since this is a steady-state process. It has been suggested that a dynamic simulation is an absolute necessity for PSVs sized for closed valve. The engineer arguing this case says that dynamic simulation can be used to justify a drastically smaller valve. There is also an argument to be made that the dynamic simulation won't be sanctioned by API so who cares.

Can you think of any other applications where a dynamic simulation is required?

What do you think?

 

[poli60]

I've experience of several dynamic simulation studies applied to refinery flare systems. Mainly, the dynamic simulations are very useful (and applied) in three cases:
1) calculations of discharge loads from systems with thermal inertia and liquid hold-up. This is the typical case for the HEATER-COLUMN systems.
2) analysis of combination of multiple loads and their variations with time. For example, in a power failure scenario you'll have various loads to the flare. If you don't consider dynamic effects, your flare shall be sized for the sum of all the peak loads. With a dynamic simulation you can model the variations of the single loads vs the time and easily see that the various peaks will be at different times. At the end, the design load for the flare will be lower. This concept is widely applied in flare revamping studies.
3) calculations of discharge loads originated by emergency depressurization systems.

 

[Chance17]

Well, external Fire with internal Chemical reaction - in today's world a dynamic simulation is almost a requirement.

I used to normally calculate both steady state and unsteady state.
The unsteady state calc was used to only report time duration required for the external fire to activate the safety valve.
If the duration was greater thn 30-minutes, I would ocssionally invoke risk base analysis to mitigate any facility deficiencies.

Today I only calculate steady state.

 

[SDwijanarko]

Dynamic simulation is used to optimize the size of relief system as it will help calculate and provide more realistic relief load. This is especially helpful for a large unit operation, LPG fractionation towers are good examples.

SD

 

[don1980]

Rocket - by default, fire sizing calculations are based on the assumption that the vessel/system is blocked in. There's nothing that requires you to apply that assumption - it's just done because (1) it's conservative, and (2) it simplifies the relief load calculation.

I think you're saying your system is open during the fire. That's OK as long as there's good justification for that. With a closed system (default assumption) fire is indeed a dynamic event, unless the system contains a single component. In most cases you're dealing with a mixture, which means the relief composition changes from the beginning to the end. Of course, the relief stream will start off high in light components and then gets heavier with time.

A dynamic simulation model will give you a clear picture of this changing composition, but in most cases I personally don't find that to be worth the time. I look at the composition and choose a conservative component, or mixture of components, to use as a sizing basis. A low Hvap is generally the worst-case for sizing purposes. You can spend a great deal of time doing a dynamic simulation, and come up with an answer that's not significantly different form the answer based on your intuition.

Risk tolerance varies from one company to another, and one individual to another. I think you have to have a very high risk tolerance to say that a dynamic symulation will greatly reduce the size of the PSV. For example, let's say that the dynamic model (changing composition) shows that a P orifice is needed in the early stage of the event and an F orifice is needed at the end. Yes, you can save a lot of money by installing the F orifice. But, to me, what you're really doing is saving money by shifting your risk tolerance, rather than saving money by applying a dynamic model.

 

[rocketscientist]

I can see a value to dynamic analysis for some chemical reactions but I have trouble believing that heater coil relief would require dynamic analysis. Temperature controlled loops seem to be the antithesis of dynamic calculations. Besides, if a dynamic analysis is necessary for a chemical reaction can it safely represent a runaway reaction? I would like to see an endorcement from the makers of H2O2 --- they're still requiring lift plates in place of PSVs or PSEs. Then, there is the question of whether the dynamic model is really effective. And, of course, it would help if DIERS or API endorced these calculations over traditional steady-state and empirically-based calculation methods. I believe simulation just took a sock in the eye over DuPont's de-classification of a vinyl floride slurry tank that resulted in the death of a welder. Simulation told the safety engineers that the slurry would never produce enough vapor to be considered dangerous. Obviously, either their analysis was not broad enough to consider the possibilities or they should have relied on empirical methods to monitor the jobsite as a safeguard; as the CSB concluded in 2010 before the accident occured. And, if this is true then are simulations that accurate in representing properties and conditions for PSV-sizing?