Page 323 - Handbook Of Multiphase Flow Assurance

322 10. Research methods in flow assurance

(2) The rotation and translation algorithms were tested for proper reorientation and
positioning of monomers using a set of discrete angles (0,45, 90°) and steps in each
dimension.
(3) The sampling bin size of the radial correlation function was optimized. The area of
produced correlation function was integrated to yield a normalized value of 100 as
required.
(4) Proper functioning of cubic and rhombic periodic boundary conditions was verified.
(5) It was shown that the starting position of a moving molecule near the fixed molecule
does not affect the resulting strength or place of adsorption by performing 8 simulations
of one water molecule moving around a fixed PVP monomer.
The following variables were also tested: position of hydrate slice; thickness of slice; length
of equilibration; attractive effect of non-bonded interactions; number of runs required for
each slice; size of the hydrate surface (1 × 1 or 2 × 2 unit cells area).
Using the simulation code
After testing the code, the simulation of water molecule movement around a fixed inhibi-
2
tor monomers was done to compare it with the results obtained using Cerius . In 10 trial runs
the sufficient equilibration “time” for the running-average energy was determined to be at
most 26,000 steps. Simulations consisted of 200,000 move attempts with radial correlation
functions analysis output every 100,000 steps.
The temperature was fixed at 273.15 K. Charges on the water molecule were set to the
DREIDING forcefield values of −0.68 on oxygen and 0.34 on each hydrogen. The resulting
dipole moment was 2.01 Debyes, which is close to the experimental value of 1.85 Debyes for
water in the gas phase. The SPC parameters for water were adopted in later simulations.
Charges on the vinylpyrrolidone monomer were calculated using the QEq charge equili-
2
bration method available in the Cerius program. The QEq charge calculation method was
proposed and validated by Rappe and Goddard (1991). Their method showed excellent
agreement of calculated dipole moments with experimental data for 20 salts and 7 simple
organic compounds like water or methane. Also comparison of the QEq method with other
2
methods (like AMBER ) showed acceptable agreement for 18 simple organic compounds and
several polypeptides.
In our work for a PVP monomer the dipole moment vector was found to lie close to the
plane of the pyrrolidone ring, pointing from the oxygen atom towards the nitrogen atom.
The following monomers were tested using the hand-written code: vinylpyrrolidone,
vinylcaprolactam, vinyl-N,N-dimethylacrylamide, vinyl-N-acrylamide, and vinyl-alcohol
(a.k.a. ethanol). It was found that the water molecule always moves close to the carbonyl
oxygen of the pyrrolidone ring. The peak of the O water -O pyrrolidone radial correlation function
was found to be between 2.9 and 3.0 Å (Fig. 10.77). Investigation of the effect of the initial
water position with respect to the PVP monomer in the simulation box was done by running
eight simulations; in each simulation the water molecule was initially placed in a different
octant. The final running- average energies were found to have a maximum standard devia-
tion of 6.2% after 200,000 steps. The OO radial distribution peak was between 2.9 and 3.0 Å
in all cases. Thus it was determined that the starting position of the water molecule in the
simulation box does not affect the results.