Underpredicted J-T Effect - Hysys 1

[ChEMatt]

Looking at an 1100 psig (well gas, free water and oil removed, compressed into a pipeline) gas stream used for fuel gas at 100 psig. In reality, the depressurization of the stream through control valves causes a nice ice block to form over the valve and nearby piping. In Hysys, the outlet temperature is 38F, nowhere near freezing.

I'm using a recent gas analysis, so I believe it to be correct (and I've got nothing else to go on, so if it's wrong there's that). Is there something I should be doing to improve the temperature depression prediction to get something more accurate to reality? I'm using the Peng Robinson EOS, standard hydrocarbon components up to "C7+", using the "saturate" extension of Hysys to saturate the stream. We're pretty close to sea level, so no significant changes in absolute pressure are needed.

Am I forgetting something?

Thanks!

-Matt

 

[Latexman]

Is the temperature of the high pressure stream accurate, or a best guess?

Good luck,
Latexman

To a ChE, the glass is always full - 1/2 air and 1/2 water.

 

[georgeverghese]

Plain vanilla versions of thermo models in Pro II and Hysis work ok when CO2 + H2S content is less than about 5% v/v or mole . For higher concentrations of these, use the SRK KS or the SRKM or the SRK KD thermo models in Pro II - not sure what the equivalent models in Hysis are. Recommendations for these may be in the Hysis online help manual.

 

[ChEMatt]

Thanks for the replies guys.

The temperature of the high pressure stream, inlet, is accurate. I don't know where the sample was taken, however, and with uninsulated pipe it could have dropped a couple degrees by the time it hits the control valve. I'll check the sensitivity of the outlet with a few degree change on the inlet.

This is sweet gas, so < 4 ppm H2S and CO2 is less than 2%, probably about 1.2% (not looking at the composition at the moment).

PR seems to be Hysys' bread and butter, but it's a great suggestion to check other correlations. I'll do that.

Any other hints or thoughts?

-Matt

 

[MortenA]

Are you sure that the gas is water saturated? Water would drop out and freeze (in the simulation) and this would "keep temperature up" - and over prediction of the MW would also lead to a higher temperature than what you observe.

 

[don1980]

Have you measured the actual temperature. We know it's less than 32, but how much less? In other words, how far off is this 38F prediction made by the EOS model? You may be expecting too much accuracy from an EOS model of a mixture. Regarding the composition, I think the key suspect is H2. Based on my experience in process modeling, I'm not ready to describe this result as being way off. Your expectation for better accuracy is reasonable if this is a pure component stream, but the non-ideality of mixtures means that you can't always expect highly precise predictions from an EOS model.

 

[georgeverghese]

An approximate method to estimate T2 would be to use figures 24-9 to 24-15 for paraffinic vapors in the chapter on Thermodynamic properties in the GPSA, knowing this is an isenthalpic letdown.

 

[wpm1]

Another factor is the sampling technique used, and location, which can create error in the analysis results. Kinda have to do that sample like a GOR reservoir analysis or garbage in equals garbage out.

 

[apetri]

I agree with last (13 may) post about sampling techniques,
also note that eng-tips forum has many threads discussing how to calculate phase equilibria for Hidrocarbons + water, a EOS with std. mixing rules may not produce accurate results,
for HC+water you may consider GERG, CPA or some other models (PRX-WSO etc),
for some comments see this thread

http://www.cheresources.com/invision/topic/17628-hydrocarbons-dew-point-in-dcs/

 

[LittleInch]

I did like the opening statement - "In Hysys, the outlet temperature is 38F, nowhere near freezing" ERRR, it's only 6F / 3.3C from freezing. You're going from 1100 psig at an unstated temperature to 100 psig with a mixture of components in there. That's a pretty big pressure drop and I think 3 C is well within the accuracy you're going to get. I don't know what Hysys clamins, but if I got within 5C of a simulation I would consider myself lucky. It must be said, most times the packages are more conservative than reality, but not always.

Any simulation tool is exactly that - a simulation - based on a set of parameters worked out and tested in laboratories on small scale test rigs. You seem to be perhaps mistaking the number of significant digits that HYSYS can spit out with accuracy. This is incorrect. I think you are expecting way too much from the simulation package.

The other possibility is that within the valve itself, there are some elements which reduce the pressure even more and hence temperature, before the pressure recovers after the orifice or cage path. Hence this internal valve temperature might be well within 3C of your calculated gas pressure on the outlet which is all that the package knows. Have you asked the control valve guys to run it through their package?? or checked the valve suppliers data sheet?? You may well find they calculate a lower min gas temp within the valve itself.

What external temperature did you assume? Once ice forms e.g. on cold night, it won't go away.

Even if free water is removed, by the sound of it you're compressing it after that so more water might drop out and be removed? Are you sure the composition and water content account for that?

Bottom line - I wouldn't waste too much time on it - a 3C margin is WAY too low. Anything less than 10C should be ringing alarm bells. Heat it up a bit more before you take the pressure down and all your problems will go away - though the pipe and valve will probably drip water.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[apetri]

while I agree with littleinch we should consider that a measured temperature is not given (the original post says "causes a nice ice block to form over the valve and nearby piping") so it seems difficult to speculate about errors (which may be much larger),
however with possible ice / hydrates formation problems could happen *inside* piping and, in general, it is preferable to adopt accurate models for these simulations (see my previous post)

 

[cosh]

really, at those conditions well gas + free water may cause several problems...