20 replies to this topic

[MelB]

Hi everyone,

 

I am developing an Excel file for relief valve sizing, and I am stuck on the fire case.

 

I have set up the calculation such that the fire heat absorption, Q, and latent heat of vaporization at STP, Hvap, are used to calculate relief load, W = Q/Hvap. Unfortunately, I think it is far more complicated than that.

 

My next thought is to find the boiling point of a given liquid at the system conditions (pressure) and use specific heat to find heat input, H, required to get to the boiling point. The relief load would then be W = Q/(H+Hvap).

 

IF that is the correct method to determine relief load, I have several issues remaining. Our systems are fairly high pressure (~1350psig) and I understand that boiling point, specific heat, and latent heat of vaporization are all dependent on pressure; however, I cannot for the life of me find data for those variables vs. pressure or equations to describe their relationships. The liquids in question would be mainly water, MDEA, and glycol.

 

I know of the Clapeyron equation; however, it appears to assume Hvap is a constant value.

 

Am I overthinking this? Or is there data available that I am overlooking?

 

Thanks in advance!

Melissa

 

[fallah]

Melissa,

 

You cannot consider sensible heat in the equation of relief load calculation for fire case; then just the latent heat should be considered. As per API 521 you can use a minimum value of 115 kj/kg for the latent heat if no accurate latent heat value is available...

[MelB]

Okay so for fire case you neglect sensible heat which is the conservative way to do it, makes sense. Still, latent heat of vaporization is dependent on pressure, correct? Is there pressure vs. latent heat correlation data available anywhere for pure liquids?

[Bobby Strain]

If you are doing serious engineering, you should have access to a good process simulator for properties. And use it to get the properties you need. Also, some of the relief valve manufacturers will give you free software so you don't need to reinvent the wheel. You should consider linking a reliable property generator to your spreadsheet so you are not burdened with so much unnecessary work.

 

Bobby

[serra]

I agree with Bobby,
if you know composition better to get reliable values for properties,

you need to solve two different problems,
1) find the initial equilibrium temperature (given pdisch from accumulation),
to find the initial equilibrium temperature in Excel I recommend this procedure

http://prodesoftware...conditions.html

2) simulate the vaporization process inside the vessel,
as light components go into vapor (to flare) remaining heavy components determine an increasing of equilibrium temperature and properties of mixture change.
To simulate the process I have a Excel page (from Prode) which
solves a series of H-P flash operations, see the HPF() Excel macro available in Prode Properties,
at each time step the procedure estimates the heat by fire and calculates the amount of vapor to discharge (mass & heat balance) which is the rigorous solution, the procedure is described in this paper useful applications of process calculators 1994,
The Excel page takes one / two seconds for solving complex (multicomponent) operations.
Prode includes also an automatic procedure (depressuring module) but I prefer trhe Excel page as it allows me to modify the different parameters as radiation, insulation factors etc.

if you have a simulator (as hysys) you can adapt the same procedure solving the H-P operation with an exchanger or similar operation.

Edited by serra, 31 July 2014 - 12:29 AM.

[fallah]

Melissa,

 

To estimate the required latent heat value you can use a proper process simulator such as HYSYS...

[serra]

in addition to the complex (but accurate) procedure discussed in my previous post
I forgot to mention that you may consider also an estimate based

on latent heat (API 521, as suggested by Fallah),

if your mixture contains, for example, mainly water (say > 98%),

you may also consider the properties of water (i.e. saturation pressure and enthalpy of vaporization).
For a mixture you can calculate Enthalpy of Vaporization (latent heat) as SU.M(Wi*HVi)
where Wi is mass fraction and HVi is latent heat (mass basis) of component i,
you may interpolate data from tables or estimate Enthalpy of Vaporization in different ways,

including differentiation of vapor pressure correlation,

see chapter 7 of The Properties of Gases and Liquids (Reid, Prausnitz, Poling) for the details

Simulators usually calculate Enthalpy of Vaporization as difference Enthalpy Vapor - Enthalpy Liquid,
final accuracy depends from the thermo models included in the software.

Edited by serra, 31 July 2014 - 05:50 AM.

[MelB]

Bobby I do "serious engineering" and yes I have access to simulators. I also have access to manufacturer provided calculation sheets and will be using them; however, those sheets only calculate orifice area needed, not relief rate due to process conditions. Quite frankly I wouldn't trust a relief valve manufacturer to accurately calculate relief scenarios. My question is the result of a decision to try and avoid using a simulator because of the number of calculations being done, and the diversity of the physical applications. Simulation takes too much time, frankly.

 

I do err on the conservative side when sizing relief valves so I am already using conservative values for other variables in my equations. My concern is that the heat of vaporization goes down significantly with increase in pressure, so using the STP values is too inaccurate for comfort. I also realize that multi-component liquids are more complex than the assumption of weighted average, so I respectfully disagree with that method serra. In my case, the liquids in our facilities are typically "pure" but I need to do calculations for several large facilities - thus my desire to do the legwork up front and try to get accurate Hvap values in a personal file rather than use a simulator for every scenario. I am willing to use correlation equations if anyone knows where I can find curves. There are curves available for hydrocarbons in the Data Book on Hydrocarbons so I know they exist.

 

I will reference The Properties of Gases and Liquids, I did not recall that discussion in that book. Poling was a college professor of mine so I may just call him directly.

 

Thank y'all!

[Bobby Strain]

You can use your process simulator for calculating fluid density as the contents heat, presuming the fluid is in the supercritical region. Combine fluid heat absorption and metal heating during the fire. Calculate the time increment and mass change to get the incremental relief requirement. Use the same logic for fluids below critical conditions. There is an old article published with the details, but I have long since lost it. The calculation of heat absorption is straightforward. And, when your learn to use Excel VBA or something better to link with the simulator, the exercise becomes quite easy. Good luck.

 

Bobby

[MelB]

Thank you! I will try using the simulator to create my own curves. That may get me what I want as well. I will also try and learn how to link to a simulator, I haven't seen it done before.

 

I appreciate your help sir!

[Bobby Strain]

Which simulator do you use?

 

Bobby

[serra]

MelB,
I have two different editions of the Properties of Gases and Fluids, Fourth Edition discusses Enthalpy of Vaporization of Pure Compounds (which was your original question) at para 7.8, page 218 and following
on fifth edition you'll find the same discussion (para 7.7)

if you know the correct values for single fractions the correct way to estimate the value for mixture at some specified t, p is

HVap(mix) = SUM(Wi*Hvapi)

for each fraction at conditions far from ideal (for example high t, p) you must keep in account interactions between different chemicals,
the usual way to estimate these values with simulators is to evaluate departure from ideal  state with a EOS such as Soave, Peng Robinson, LKP etc.
the main problem with these models is that they may predict wrong values for liquid densities and (as conseguence) enthalpies,
you may find a detailed discussion in The Properties of Gases and Liquids.
To reduce these errors you may prefer extended models with specific parameters calculated for each chemical in database to fit experimental values of density and latent heat,
see this thread for additional information on extended three parameter models

http://www.cheresour...-peng-robinson/

if you can discuss the matter with prof. Poling that will be (in my personal opinion) certainly useful,

I do these calc's quite regularly in Excel,
for thermodynamics (with Excel or Matlab) I prefer a process library (Prode Properties)

over a simulator mainly because a process library is more immediate

(i.e. you have macros to solve everything without coding)
however a simulator (Hysys, ProII, Promax etc.) should produce equivalent results

and many engineers do prefer a graphical approach,
anyway you can adapt the rigorous procedure and other methods discussed in my previous posts to any software.

[MelB]

Thank you serra! I may give Dr. Poling a call he is usually very willing to help. I am with you, adapting these calculations to Excel simplifies life. I am glad to know that it is not entirely unacceptable, and in fact it is recommended, to use a weighted average. Though I'm sure there are complex interactions it is impractical to do a detailed analysis of small impurities, particularly if simulation is not accurate either.

 

Bobby I am provided ProMax by my company, which I am not really fond of but we work with what we have. I have worked with HYSYS in the past and prefer it, but unfortunately I no longer have licensing to use it.

[Bobby Strain]

ProMax should have all the OLE capability of HYSYS, maybe better to communicate with Excel. I only use HYSYS.

 

Bobby

[serra]

MelB,
I agree with you that adapting these calculations to Excel simplifies life,
in my opinion a simulator and graphics are useful for converging large networks
while to simulate simple plants or single equipments Excel (or Matlab or equivalent tools) can give additional flexibility

Bobby,
yes you can utilize OLE to link Excel but in my experience OLE is not competitive (in terms of speed, reliability, flexibility etc.) with a direct plug in,
I undestand that one has to work with what he has but there are points to consider,

for example there are free versions

(I know Prode Properties but there may be others)

which one can download :-))

[moein_omg]

I had recently stumbled across such a problem. I used the following procedure:

 

1- for the fire case I assumed that the density (average density considering all phases)  is constant as the equipment or piping section is isolated.

 

2- having the relief pressure (P_set=1.21*design pressure), one can easily calculate the relief temperature using a thermodynamic simulation software such as hysys or by developing a computational code taking advantage of the constant density. To clarify, in hysys, you can use adjust function to manipulate relief temperature to satisfy the assumption of constant density or use an iterative algorithm to calculate it in your computational code.

 

3. This (temperature and pressure) is the point where relief starts you must also be aware that the process is far more complex and the composition changes as time goes on. A complete dynamic simulation can be helpful to evaluate different cases. You can use hysys depressurizing utility to do such a dynamic simulation

 

However, in my personal point of view, such fire case safety valves and the relevant scenarios are usually useless especially dealing with gaseous streams as relief temperatures are usually far more than what is practical. It is especially valid for cases where design pressure is dictated by something apart from operating pressure.

 

I hope it would be of some help.

Edited by moein_omg, 05 August 2014 - 07:35 AM.

[serra]

moein_omg,

thanks for the contribute,
if I understand correctly your post to find the initial temperature
you suggest the same procedure
discussed in my post nr. 5 which does utilize (in Excel with Prode Properties)
a single macro = VPF() for solving immediately the flash operation
at specified Volume and pressure,
with the simulator you need to iterate so things are a bit more complex
but results should be equivalent,

after initial step I would suggest to continue with a series of H-P flash operations

(see my post #5)
 

[moein_omg]

Thanks serra;

Actually the procedure that you adopted is more complete and with less simplifying assumption. Do you follow any special reference for that? If so can you kindly introduce it?

One more question, does the procedure for multiphase phase cases results in intermittent relieving scenarios?

[serra]

moein_omg,

the VBA code includes as reference "useful applications of process calculators" Chemputers Europe, 1994.
The code itself is simple being based on a series of predefined flash operations which are available in Prode Properties, see post #5 for the description of the procedure,
I got the code with the depressuring module extension, I prefer the VBA code (over the depressuring utility available in Prode) since I can adapt the procedure when required.
I think you can create something similar with another simulator (say Hysys, ProII etc.) in case the module is not available,
finally, about your last question,

in my experience convergence for H-P flash operations (and other operations available in Prode) is generally stable in most cases,
I have not access to other simulators and I can't say how Prode compares,
but you may test yourself.

[Bhavinkumar]

For a single component it is easy to know the latent heat of vaporization as posted earlier. However for multicomponent it is not that much straight forward. For multi-component systems, the vaporization of the liquid initially in the vessel at the start of the fire proceeds as a “batch distillation” in which the temperature, vapor flow rate and physical properties of the vapor and liquid in equilibrium with each other change continuously as the vaporization proceeds. The peak relief load may or may not coincide with the start of the fire. In general, such systems require a time-dependent analysis to determine the required relief area and the corresponding relief rate. The following approach is suggested:

 

Assume that all vapor and liquid inflows into and outflows from the vessel (other than the fire relief load) have stopped. Using the composition of the residual liquid inventory in the vessel, perform a bubble point flash at the accumulated pressure. In doing this flash, the flow rate of the feed stream to the flash can be set at any arbitrary value. For convenience, it is suggested that the mass flow rate be set numerically equal to the mass inventory of liquid initially in the vessel or 1000 units of mass flow rate (kg/s). Flash the liquid from the preceding flash at constant pressure and the weight percent vaporized equal to 5%. Divide the heat duty calculated for this flash by the mass flow rate of vapor generated. The result is the heat absorbed per unit mass of vapor generated, λ. NOTE THAT, IN GENERAL, THIS VALUE WILL NOT EQUAL THE LATENT HEAT OF VAPORIZATION. In fact, the value thus calculated will generally exceed the latent heat of vaporization, especially in the case of wide boiling mixtures. The reason is that a significant portion of the heat absorbed goes into raising the temperature of the system (most of which is residual liquid at this point) to the equilibrium temperature of the flash (i.e. sensible heat).

 

Use  λ calculated by above method in API-521 formula for releiving flow for fire case.

[namvuvu29]

hj MelB,

 

you can refer my method to apply for your case.

 

Best Regard,

Hai Nam

Attached Files

•  Cal. fire case PSV in two phase.pdf   581.56KB   556 downloads

Edited by namvuvu29, 02 December 2014 - 11:33 PM.