A very robust and easy to implement interval approach to
solve VLE, LLE and VLLE flash problems
Hamid Mehdizadeh, Hallvard F.
Svendsen, Tore Haug-Warberg
Norwegian
Flash calculation problems are one of the most frequently
occurring problems in chemical engineering and lots of effort has been put into
solving its challenges. Since the early work of Rachford and Rice[1] (RR) different methods have been
suggested over the decades[2-6]. But, still, this problem is subject of
interest where research effort is focused[7].
Different aspects of flash calculation procedures are studied like robustness[6], speed[8] and
number of phases[9].
Equi-fugacity is the necessary condition for the
phase equilibrium, while a minimum of Gibbs free energy can be considered a
sufficient condition. A successive substitution (SS) approach used in the RR
method decreases the Gibbs free energy in each iteration, with some exceptions[10]. SS method consists of two loops. In the
internal loop the phase fraction is calculated and the external loop iterates
on Ki (distribution
coefficients). The external loop could always be convergent using damping factor[10], but there are problems associated with solving
the internal loop as the vapor fractions could obtain negative values [11, 12]. Some other difficulties are
discussed in Veeranna et
al.[13].
To get rid of difficulties in the calculation of phase
fraction in the inner loop an Interval-Newton method is used. Based on the this
method, all roots of a function in a pre-specified interval could be calculated
with mathematical proof, with some exceptions[14].
Interval calculation introduced by R.E. Moore in 60's, was first developed to
estimate rounding error[14]. This method is able to
predict the range of change of f(x) when x changes in the specified interval of
[a,b]. For this work, a program based on the INTLAB[15] package in MATLAB environment is used.
The function that should be solved for 2 phase flash is:
Where b is the molar fraction of the
reference phase. As the size of b is (NP-1), so the size of the problem that is to
be solved by interval analysis will not be too big and could be done with a
small computational effort. In the outer loop both equations of state and
activity coefficient models could be used. This feature makes the model
flexible and easy to use in different approaches.
Heidemann[10] describes
that with the SS method in each iteration the Gibbs free energy will decrease,
but there will be some cases with instability. To prevent instability and also
to speed up the convergence, an adaptive damping factor is used.
The prepared model showed good capabilities in prediction
of VLE, LLE and VLLE. For VLE,
different hydrocarbon and non-hydrocarbon mixtures, in a wide range of
temperatures and pressures were examined and good prediction was achieved.
Binary interaction parameters were calculated from a group contribution
approach developed by Jaubert[16] and his
coworkers in a series of papers and further expanded by Ghadrdan [17] to new groups. Three different mixtures
(nC10-CH4, CO-nC8, H2-nC6) were validated using this method. For the first
mixture the model is predictive up to 24MPa which was the limit of available
experimental data. Considering the big difference between the size of molecules
and also the very high pressure reflects the performance of the model in
reasonable computation time. All results were calculated without any additional
data. Distribution values were estimated from the Watson equation. For VLLE cases, a mixture that is considered in
CO2 flooding and EOR research was examined. For LLE,
some different mixtures were investigated using the extended UNIQUAC framework
as presented by Thomson.[18], but for non-electrolyte
systems.
Acknowledgements: Financial support from Aker Clean Carbon and the Research Council of Norway through the CLIMIT program SOLVit project is greatly acknowledged
References:
1. Rachford,
H.H. and J.D. Rice, Procedure
for Use of Electronic Digital Computers in Calculating Flash Vaporization
Hydrocarbon Equilibrium. 1952.
2.
3. Michelsen, M.L., The
isothermal flash problem.
4. Michelsen, M.L., The
isothermal flash problem. Part II. Phase-split calculation. Fluid Phase Equilibria, 1982. 9(1):
p. 21-40.
5. Parekh, V.S.
and P.M. Mathias, Efficient
flash calculations for chemical process design -- extension of the Boston-Britt
"Inside-out" flash algorithm to extreme conditions and new flash
types. Computers
& Chemical Engineering, 1998. 22(10):
p. 1371-1380.
6. Li, Y. and R.T.
Johns, A
Rapid and Robust Method To Replace Rachford-Rice in
Flash Calculations, in SPE Reservoir Simulation Symposium. 2007:
7. Nichita, D.V. and A. Graciaa, A new reduction method for phase
equilibrium calculations. Fluid Phase Equilibria, 2011. 302(1-2): p. 226-233.
8. Michelsen, M.L., Speeding
up the two-phase PT-flash, with applications for calculation of miscible
displacement. Fluid Phase Equilibria, 1998. 143(1-2): p. 1-12.
9. Fraces, C., D.V. Voskov, and H.A. Tchelepi, A
New Method for Thermodynamic Equilibrium Computation of Systems With an
Arbitrary Number of Phases, in SPE Reservoir Simulation Symposium. 2009: The
Woodlands,
10. Heidemann,
R.A. and M.L. Michelsen, Instability of Successive Substitution. Industrial &
Engineering Chemistry Research, 1995. 34(3):
p. 958-966.
11. Michelsen, M.L., Calculation
of multiphase equilibrium. Computers & Chemical Engineering, 1994. 18(7): p. 545-550.
12. Leibovici,
C.F. and D.V. Nichita, A new look at multiphase Rachford-Rice equations for negative flashes. Fluid Phase Equilibria, 2008. 267(2):
p. 127-132.
13. Veeranna,
D., et al., An algorithm for
flash calculations using an equation of state. Computers &
Chemical Engineering, 1987. 11(5):
p. 489-496.
14. Hansen, E.R.
and G.W. Walster, Global optimization using interval
analysis. 2004,
15. Rump, S.M., INTLAB
- INTerval LABoratory,
in Developments in Reliable
Computing, T. Csendes, Editor. 1999, Kluwer Academic Publishers:
16. Privat, R.,
J.-N. Jaubert, and F. Mutelet, Addition of the sulfhydryl
group (-SH) to the PPR78 model (predictive 1978, Peng-Robinson EOS with temperature dependent kij calculated through a group contribution method). The Journal of
Chemical Thermodynamics, 2008. 40(9):
p. 1331-1341.
17. M. Ghadrdan, 2011, Personal communication
18. Thomsen, K., P. Rasmussen, and R. Gani, Correlation
and prediction of thermal properties and phase behaviour
for a class of aqueous electrolyte systems. Chemical Engineering
Science, 1996. 51(14):
p. 3675-3683.
See more of this Group/Topical: Engineering Sciences and Fundamentals