Latent Heat in HYSYS for multicomponent mixture 4

[ErimusOilnGas]

Hi I am trying to get a value for latent heat of a mixture. I require the latent heat to use in W=Q/L to give me a required relief rate.

I have a few questions which noone can really seem to answer :

1 - If i copy my stream and set vap fract to 0.01, the mass heat of vap displayed in the properties tab for that stream. Is this the latent heat for the mixture or just the lightest component?

2 - In thinking method 1 was not correct I copied the stream set it to operating conditions and then put it through a heat exchanger (as is the case for vapour expansion and relief)I set my outlet to the relieving pressure . Is this the correct methos ... it give a latent heat for the outlet stream which is in the right ball park but im not sure as when i manually change the vap frac of the outlet this doesnt effect the latent heat which i would expect it to do as it is changing the temperature and as such the points at which my mixture will vaporize off.

Please ask if i have not included enough information.

Help much appreciated.

Cheers

Jonny

 

[apetri]

I am not an expert with your software however almost all simulators allow to calculate latent heat as difference between enthalpy of gas and enthalpy of liquid,
for example with Prode and Excel to calculate latent heat I write in a cell

=StrGH(1)-StrGL(1)

which returns the desired value,
since your simulator exports methods in Excel probably you have a similar option

 

[MortenA]

apetri - carefull now

For calculating the latent heat e.g. for a safety valve calculation you DONT want to "evaporate" all of the liquid - because this will give you a much higher latent heat than the INITIAL latent heat (the heavier components have a higher latent heat - and if water is a part of the stream it will shoot thhrough the roof).

In HYSYS take a copy of your stream. Set the pressure to your relief pressure (+21% if the is an API 520 sizing) and set the vapour fration to zero - this will calculate the boiling temperature.

Now and a heater unit operation and add "enough" heat to vaporize a small fraction. This should not be .00000001 because thats too harsh, but maybe something like 0.001-0.01.

Then do the math:

Q_lat (KJ/KG)=Q (added heat)/mass evaporated (or use mol if you are looking fo molar latent heat).

 

[apetri]

MortenA, carefull now

given a gas and liquid composition at specified T,P, (supposing we are in two phase area)
if you subtract the enthalpy of liquid phase from that of vapor phase you **DO NOT** "evaporate" all of the liquid,
but only the vapor phase with the composition calculated by your simulator (at specified T,P),

from that point of view the two procedures are equivalent.

 

[ErimusOilnGas]

Should the latent heat value (mass heat of vap) not change as I alter the vap fraction ? Surely the latent heat value should change as the vap fract increase(temp therefore increases) heavier end start to vapourize off?

Is the latent heat for the mixture or simply for the light ends? at bubble point 0.01 vap frac?

Cheers

Jonny

 

[apetri]

for a multicomponent mixture you can define a latent heat for the vapor or (incipient vapor) fraction,
to calculate the latent heat for the mixture SUM(Wi*Hvi) where Wi is the (weight fraction) composition of gas fraction and Hvi the latent heat of each component,
according the above if the composition changes (when vaporizing) (most probably) also latent heat will change,
I do not know your software, the procedure described by MorternA seems (to me) quite complex but that is a personal comment being familiar with another software (Prode) which allows to calculate these properties directly in Excel

 

[MortenA]

I still dont agree with apetri.

At the operating P&T assuming that we have a two phase (or 3 phase) mix in a vessel then at a fire what happens is:

1) Pressure start to increase due to the bubble point P increaese
2) Pressure meeets SP of your valve, this is not so important because you only want to size for your max rfelief case (for API 520 sizing at SP+21%)
3) So now the temperature has increased so much that you have a vapour pressure= SP+21%
4) You are NOT interested in what the "average" latent heat of evaporating all the light components have been so far - what you want to know is: What is the laten heat required to evaporate the "next" fraction.
5) Therefore you do as i earlier stated: add a "little" heat and see how much this evaporates (mol og mass) and then calculate the equivalent laten heat mass or mol.

This can then be used for sizing the PSV in accordance with api

Again i think that apetri "oversimplifies" it when he writes "to calculate the latent heat for the mixture SUM(Wi*Hvi) where Wi is the (weight fraction) composition of gas fraction and Hvi the latent heat of each component" bacause Hvi will be pressure/temperature dependant AND its not what the concentration of the current vapour phase is - its what the composition of the next fraction that evaporates off is.

I have worked with HYSYS for +15 years and has found that the methode that describe is fast and gives accurate results (OK i have rarely had the chance to benchmark these results (truth: never))

Best regards

Morten

 

[apetri]

Morten,
I believe that for a incipient vapor fraction (see my previous post) the correct solution is
Hv = SUM(Wi*Hvi)
where Wi is the (weight fraction) composition of gas fraction and Hvi the latent heat of each component
Hvi = Hvgi-Hvli (Hvgi enthalpy of vapor, Hvli enthalpy of liquid)
this is a rigorous result, not an estimate as you suggested, see all the literature including the Properties of Gases and Liquids by

Prausnitz, Poling, Reid.

As you correctly noted the main problem when estimating latent heat for a wide boling mixture with the purpose to size a PSV is the

definition of some average or maximum value,
Your procedure has several limits, you calculate a "istantaneous" value (equivalent to Hv = SUM(Wi*Hvi) discussed above) and you assume

vapor-liquid equilibria when discharging, which could not be true.
There are different alternatives, for example one could estimate the initial condition (when PSV opens) and from that calculate some average latent heat (or some other values required to size the PSV) .
Consider a simple case, a HP vapor-liquid separator with operating conditions 50Bara 15C
the separator is protected by a PSV with set point (opens at) 100 Bara

composition (mol)
C1 0.7
C2 0.04
c3 0.07
c4 0.12
cO2 0.05
H2O 0.02

with model Peng Robinson, standard correlation and BIPs from database the phase envelope is that included (blue line on the right being

the water dew point line)

at operating conditions of 50Bara 15C the calculated compositions (vapor-liquid-liquid) have been included

To estimate the conditions when PSV opens one can solve a Volume-Pressure flash operation,
we know the volume at operating conditions (50Bara 15C) which is about 0.0133 m3/kg, the initial pressure (50 Bara) and the pressure

(100 Bara) when the PSV opens, We assume that volume of vessel (and fluid) is constant.
Solving the V-P flah operation returns a temperature (@100Bara) of 168C (much higher than Bubble point temperature) with the fluid all in

vapor state.
Knowing the conditions at 100 Bara one can calculate the difference of enthalpy (@100Bara 168C - @50Bara 50C) and from that estimate the

values required to size the PSV

Working with Excel and Prode Properties I can solve quickly these calc's with a very limited effort (just two macros in Excel cells)
and compare results from different procedures (for example that which you suggested),
this is in my opinion a important advantage.

 

[MikeClay]

Apetri:

The method that you describe will end up treating the wide boiling mixture as a single component with a heat of vaporization in the range of 50-150 BTU/lb. However, in order to boil off any vapor, the temperature must rise. In wide boiling mixtures, the sensible heat can account for 80-90% of the total heat input. Morton's method takes this effect into account. I've seen instantaneous heats of vaporization as high as 2000 BTU/lb. If a dynamic model is run, the required relief area will be similar to Morton's analysis. It should also be noted that the maximum relief area is seldom the initial relief area. In all of the dynamic models that I have run with crude oil, the dynamic model gives a smaller relief valve size than just using the Hysys/Unisim mass heat of vaporization.

--Mike--

 

[delittle]

Mike,
the method described by apetri Hv = SUM(Wi*Hvi) where Wi is the incipient vapor fraction (see above posts) should returns the same values as that described by Morten, they both return "istantaneous" values (as mentioned by apetri),
You seem to be inferring your conclusion from the description of the procedure for estimating t,p,w conditions where PSV opens, that has a different scope.

 

[MikeClay]

The two methods will create latent heat of vaporization (HoV) that differ by a factor of 5 to 10. The individual HoV will be in the range of 50 to 150 BTU/lb (see the chart in API 521). If you sum the weight fractions times the individual HoV, the number will still be in the range of 50 to 150 BTU/lb. The Method that Morton describes will typically give a heat of vaporization of 300 to 1000 BTU/lb. I know these values seem high, but for a wide boiling crude, a lot of the heat is absorbed as sensible heat.

--Mike--

 

[delittle]

Mike,
the original post did ask a method for calculating the latent heat of a mixture

>I am trying to get a value for latent heat of a mixture.

and Hv = SUM(Wi*Hvi) where Wi is the incipient vapor fraction is a correct answer

in my opinion the method proposed by Morten

>Set the pressure to your relief pressure (+21% if the is an API 520 sizing) and set the vapour fration to zero - this will calculate the >boiling temperature.
>Now and a heater unit operation and add "enough" heat to vaporize a small fraction. This should not be .00000001 because thats too >harsh, but maybe something like 0.001-0.01.
>Then do the math:
>Q_lat (KJ/KG)=Q (added heat)/mass evaporated (or use mol if you are looking fo molar latent heat).

does consider a initial value (with sensible + latent) which for wide boiling mixtures may be very different from some average or minimum value (for latent heat) and as result you can undersize the PSV,
also the method may be not applicable in many cases (see the example provided by apetri where the mixture is all in vapor phase when the PSV opens),
considering these limits it would be preferable a method to estimate the initial conditions (as that proposed by apetri) or run a dynamic simulation (which should produce equivalent results).

With a fire case one should evaluate the whole process to evidence high temperatures and possible failures of vessel (keep in mind that we are discharging at constant pressure not depressuring),
that is a very important point to consider.

 

[don1980]

It's important to recognize that no matter how you size the PSV, it's a short term and very imperfect layer of protection from fire exposure. Whenever you have a mixture, especially one with a wide boiling range, there's just no single sizing basis that's objectively correct.

In a lot of cases the PSV isn't going to adequately protect the vessel regardless of the PSV size. I'm referring to cases such as:
[ul]vapor filled vessels
vessels containing supercritical fluid
vessels containing a high boiling liquid
vessels with a lot of unwetted surface area
vessels which will quickly become empty of liquid[/ul]

For those vessels we ought to be focused on other layers of protection because the PSV (no matter the size) isn't going to provide meaningful protection. In such cases consider one or any combination of the following: (1) fire resistant insulation, (2) water spray, (3) auto depressurization.

Most hydrocarbons have a Hvap in the range of 100-180 btu/lb. Once the PSV opens the mixture composition will inevitable change with time. Personally, I don't see much value in spending a lot of time selecting the Hvap, as long as I've selected an Hvap that's reasonable for that particular mixture.

To strongly argue for one value over another you have to know how long the fire will last, but of course we don't know how long it will last, and if it last too long then the vessel fails regardless of which number we choose.

 

[MortenA]

Im on the "client side" but has also been on the consultant side. If a consultant came to me I would not accept the methodology suggested by aperitif unless it did not cause me to choose a large than required psv such as I think there is a risk off.

 

[apetri]

I agree with Delittle

"considering these limits it would be preferable a method to estimate the initial conditions (as that proposed by apetri) or run a dynamic simulation (which should produce equivalent results)."

and don1980

"Most hydrocarbons have a Hvap in the range of 100-180 btu/lb. Once the PSV opens the mixture composition will inevitable change with time. Personally, I don't see much value in spending a lot of time selecting the Hvap, as long as I've selected an Hvap that's reasonable for that particular mixture."

Morten,
you do not share a numerical case and I am unable to compare the values, however there is this post from MikeClay

"The two methods will create latent heat of vaporization (HoV) that differ by a factor of 5 to 10. The individual HoV will be in the range of 50 to 150 BTU/lb (see the chart in API 521). If you sum the weight fractions times the individual HoV, the number will still be in the range of 50 to 150 BTU/lb. The Method that Morton describes will typically give a heat of vaporization of 300 to 1000 BTU/lb. I know these values seem high, but for a wide boiling crude, a lot of the heat is absorbed as sensible heat"

according that your procedure calculates a higher latent heat and (consequently) a lower mass flux and lower discharging area, which may be not acceptable,
in my opinion if you do not wish to utilize a procedure for estimating initial conditions and from those some average / maximum value for latent heat as I do in my example then you may run a dynamic simulation and see what happens,

but do not forget these comments from delittle

"With a fire case one should evaluate the whole process to evidence high temperatures and possible failures of vessel (keep in mind that we are discharging at constant pressure not depressuring), that is a very important point to consider."

and don1980
"To strongly argue for one value over another you have to know how long the fire will last, but of course we don't know how long it will last, and if it last too long then the vessel fails regardless of which number we choose"

 

[ErimusOilnGas]

Guys, firstly thank you all for your input into this thread.

TO give you abit more background, I work for the client and am currectly reviewing PSV calcs which have been completed by an outside consultant, namely looking at those which have highlighted anomalies where a suggestion that a valve is undersized has been made. Myjob is essentially to sense check there assumptions.

In this case the consulting party has stated the boiling range for the mixture and stated they will use the lowest latent heat as this will be the worst case relief from W = Q/L. This worst case however is something which we see in industry too many times and where it can be a more accurate case should be provided otherwise the valve ends up being oversized and PSV chatter occuring.
They had also assumed a component mixture where I am now taking average smaple reading to try and give the best case relief.

The 'Dynamic Modelling' suggstion is probably the best option however we do not have that at our disposal. Therefore it becomes difficult to really choose a vap frac at which to take the latent heat .. as previously stated this will indeed change as the temperature increases.