9 replies to this topic

[cxl]

Hi all, I have a question here with regards to joule-thomson cooling effect. I have a gas-condensate of known fluid composition. How do I use PVTsim to find out the cooled down temperature after the fluid experiences pressure drop. I have the inlet and outlet pressures, and inlet temperature condition.

Thank you so much in advance!

[Zauberberg]

Joule-Thomson effect applies for pure gases, not mixtures of gas (vapor) and liquid. Temperature changes for gases can be calculated based on the value of J-T coefficient at given temperature range.

Without process simulator, calculation is much more difficult and it is iterative in nature. Following the basic principle of valve flash calculation (isenthalpic process), you would assume valve outlet temperature, then calculate vapor-liquid compositions, and then calculate enthalpies of both phases and compare with the inlet stream. Iterations are repeated till you match both values.

[cxl]

Thanks for your reply!

I understand that the J-T effect is for pure gases only, so could I then do a flash under standard conditions in PVTsim for the gas-condensate fluid, and capture the vapor composition, input the vapor as another fluid, then do another flash for this vapor under inlet pressure and temperature conditions, and use the J-T coefficient to calculate the temperature drop, and hence the final cooled down temperature?

I also have the OLGA software, is it possible to use that to investigate the J-T cooling effect?

[PaoloPemi]

Joule-Thomson is defined as dt/dp (constant H). I don't know your software, but several simulators calculate these values for gas and liquid phase. For two-phase with my software (Properties, see prode.com), to get a idea of dt/dp I use finite difference with H-P flash operation. However, these values are strongly related to VLE equilibria and the resulting function can be very nonlinear.

Paolo

[cxl]

Thanks for your reply Zauberberg!

I understand that the J-T effect is for pure gases only, so could I then do a flash under standard conditions in PVTsim for the gas-condensate fluid, and capture the vapor composition, input the vapor as another fluid, then do another flash for this vapor under inlet pressure and temperature conditions, and use the J-T coefficient to calculate the temperature drop, and hence the final cooled down temperature?

I also have the OLGA software. Is it possible to use that to investigate the J-T cooling effect?

[marchem]

to get the exact value you should add the contribute of phase change, see for example " Calculation of Joule–Thomson inversion curves for two-phase mixtures" by Nichita and Leibovici

[PaoloPemi]

to get the exact value you should add the contribute of phase change, see for example " Calculation of Joule–Thomson inversion curves for two-phase mixtures" by Nichita and Leibovici

interesting work, thank you

[sheiko]

Just to add my 2-cents,

The Joule-Thomson effect doesn't always imply the cooling of the gas upon expansion. Some fluids warm (negative Joule-Thomson coefficient). Hydrogen is the more common example.

Zauberberg,
I have read in the article untitled "Working with fluids that warm upon expansion" by John J. Carroll (issued in CHE magazine - September 1999) that Joule-Thomson effect may be encountered with mixtures (gas or liquid). The author shows, as an example, how the temperature of a black oil increase with decreasing pressure.

Edited by sheiko, 17 October 2010 - 12:27 PM.

[fallah]

Joule-Thomson effect applies for pure gases, not mixtures of gas (vapor) and liquid.

Hi Zauberberg,

As far as i know Joule-Thomson effect applies for gaseous mixtures as well.Would you please submit a valid reference to show your above statement.

Regards

[PaoloPemi]

Joule-Thomson is defined as dt/dp (constant H) , you can calculate this value for both gas & liquid phase starting from cp and dV/dT (const P), some simulators and tools as NIST Refprop or Prode Properties have analytical derivatives and the returned values are numerically very accurate, of course the real accuracy depends from model.