Page 267 - Handbook Of Multiphase Flow Assurance

266 10. Research methods in flow assurance

Verification of water models
In order to verify that water properties are represented adequately by the model, com-
parison of the three models available in SYBYL against the experimental data was done.
These models are SPC (simple point charge), TIP3p (transferable intermolecular potential
3 point charge) and the proprietary Tripos model. All these models are the three-center
models of water. Radial distribution function was chosen as the check parameters for the
comparison. Default parameters preset in SYBYIT were used for these simulations. These
include the 0.8 nm interaction cutoff and 10 nm Ewald summation of electrostatic interac-
tions radii. All atoms in water molecules had partial charges on them. Charges for the SPC
and TIP3p models were set as the default ones, and for the Tripos model charges were set
equal to +0.41 on each hydrogen and − 0.82 on oxygen as in the SPC model. As the result
of this part, the SPC water model was chosen as the best to simulate the structure of real
water.
The importance of having charges on the atoms of water cannot be underestimated. A
run was performed by mistake for the Tripos water model without charges. A change of the
radial distribution function was very significant compared to the correct simulation and the
experimental data.

Procedure of water models verification
A lattice of 216 (6 × 6 × 6) water molecules was set up for each model with the density
of 1.0 g/cc. Periodic boundary conditions were applied. Charges on the SPC water mole-
cules were set as precomputed (+0.41 on each hydrogen and −0.82 on oxygen). Values of
the charges are the fractions of the electron charge (4.8 × 10w esu). The physical signifi-
cance of the charges in an electroneutral molecule is the distribution of an electron cloud
between the atoms of this molecule. In every simulation the system potential energy
was minimized prior to the run. Conjugate gradient minimization routine of SYBYLF
was used.
The initial part of each run was equilibration of the water in the periodic box. It consisted
of four stages:
(1) 1000 time steps in microcanonical NVE ensemble (constant number of particles N,
constant volume V, and constant energy E) to equilibrate kinetic and potential energies
of water and to melt the water lattice.
(2) 1000 time steps in canonical NTV ensemble (constant N, constant temperature T,
constant V) to set the temperature to 300 K.
(3) 1000 time steps in microcanonical NVE ensemble.
(4) The last part was the data collection part of 10,000 time steps in NVE ensemble. Each
time step equals 1 fs or 10 −15 s. Default parameters preset in SYBYL’ were used for these
simulations.

Each simulation in this part used the previous starting velocities of atoms. This means that
the initial kinetic energy or the temperature of the system had to equilibrate from the initially
minimized value (effectively ~0 K). The implementation of molecular dynamics in SYBYL al-
lows a choice of force fields (as it is also done in the SYBYL energy minimization facility) and
uses the Verlet (1967), also known as the Leapfrog method, for the integration of the equations
of motion. In each simulation the TRIPOS forcefield was used.