Page 297 - Handbook Of Multiphase Flow Assurance

296 10. Research methods in flow assurance

This can be easily rationalized by assuming that growth involves the consecutive addition
of growth planes of molecules. If energetic effects are discounted, the ease of adding a plane
is proportional to its thickness. Thus, a thinner plane grows faster and has a larger center-
to- face distance. Donnay and Harker (1937) refined this approach by developing rules that
related the crystal symmetry to the possible growth planes.
2
The Cerius program was used to report correctly the shapes of sI and sII single hydrate
crystals as regular octahedron and rhombic dodecahedron, respectively. These shapes were
also observed experimentally (Larsen et al., 1996; Makogon et al., 1997). sH hydrate crystal
was predicted to grow as a hexagonal prism. All calculated shapes were later confirmed by
the work of King (Smelik and King, 1997).

Inhibition of hydrate growth
Effects of additives on crystal growth from aqueous solutions have been studied previ-
ously for calcium sulphate (McCartney and Alexander, 1958), adipic acid (Colville, 1958),
ice I h (Knight et al., 1991), glycine (Li et al., 1994). However, such experiments for clathrate
hydrates have started just recently for sII (Makogon et al., 1997), and sI (Larsen et al., 1996)
hydrates. Computer simulations have predated experimental work in this area.
In 1993 Edwards has presented the results obtained with the Cerius and CHARMm soft-
ware for the adsorption of polar fish glycopeptides on hydrate and ice (Edwards, 1994). He
indicated that the optimum spacing for the polypeptide adsorption was on {100} face of sI
hydrate in a 〈110〉 direction. He reported that 15 hydrogen bonds formed from aminoacid side
chains and some carboxyl groups to the hydrate surface. This number is qualitatively higher
than the fraction of hydrogen bonds measured between poly(methyl methacrylate) carbonyl
groups and hydrogen terminated {100} silanol (silicon oxide) surface measured as 0.09–0.13
(Zazzera et al., 1993).
Intramolecular motions of hydrophilic polymers like poly(dimethylacrylamide)
(PNNDMAM) in aqueous solution are measured to be of the timescale which can be mod-
eled by computers, ca. 3.4 ns (Soutar et al., 1996). Such simulation using MD would be long.
Usually adsorption of flexible polymer chains on surfaces is modeled by self-avoiding Monte
Carlo technique using the cubic lattice for positioning polymer links in solution modeled by
vacuum (Konstadinidis et al., 1992; Zhan et al., 1993; Zajac and Chakrabarti, 1994). Such sim-
ulations are suitable for calculation of the surface coverage with polymer segments and radii
of gyration. Cubic lattice allows to simulate polymer only as a chain of segments with a dis-
crete attractive, repulsive or neutral potential (e.g., -kT, 0, 0.5 kT, kT). No atomic information
can be modeled using the lattice model.
However, more complicated methods involving chain rotations and cooperative motions
have to be used if polymer links are composed of multiple atoms (e.g., pyrrolidone). A pivot
algorithm for polymer chain rotations was reintroduced by Madras and Sokal (1988). The
algorithm makes very bold changes in polymer backbone conformation. Many such moves
get rejected because polymer segments overlap. However, the few moves which are accepted
produce such changes in conformation which would have taken many regular movements
of polymer segments on the cubic lattice. MC simulations involving pivot moves were used
to study adsorption of monomers, homopolymers and copolymers on interfaces (Clancy and
Webber, 1993, 1997).