Yun Hee Jang, Mario Blanco, William A. Goddard, III
MSC, Beckman Institute, Caltech
Augustin J. Colussi, Michael R. Hoffmann
Department of Chemistry and Chemical Engineering, Caltech
Yongchun Tang, Bob Carlson, Huey-jyh Chen, Jefferson Creek
Chevron Petroleum Technology Co.
cold
sea
water
cold
sea
water
hot
oil
wax
Wax: Aggregates
of heavy n-alkanes
at low temperature
 pipe blocking
oil production pipe wall
cold
sea
water
cold
sea
water
Comb-like wax inhibitor
Wax inhibitor
(comb-like polymer):
No established
mechanism of action.
Wax Formation
Liquid

Amorphous solid

Ordered crystal

Further growth

Adsorption on pipe
Wax Inhibition
(1) Sequestering mechanism
long alkanes in oil selectively partition toward the inhibitors
making them less available to nucleate a wax crystal
(2) Incorporation-perturbation mechanism
inhibitors partition from the oil into amorphous wax ("soft wax")
slowing down the crystallization of soft wax to form "hard wax”
(3) Wax crystal adsorption mechanism
adsorption of inhibitors on initial wax nuclei or growing wax crystals
inhibits further wax growth
(4) Pipeline adsorption mechanism
adsorption of inhibitors on the pipe wall provides an irregular surface
that interferes with adsorption of wax to form crystals
Objective of this work: Establish mechanism by investigating each of them
Hydrocarbons and long alkyl sidechains
United atom model (SKS) (Siepmann, Karaborni and Smit, Nature, 365, 330 (1993))
Stretching from AMBER with r0=1.54 Å from SKS
Acrylate backbones (around -COO-)
VdW: OPLS (Briggs, Nguyen and Jorgensen, J. Phys. Chem. 95, 315 (1991))
Charge: HF/6-31G** calculation
Torsion: fitted to HF/6-31G** torsion energy curve for model systems
Stretching/bending/inversion: AMBER (r0,0 from OPLS)
Styrene backbones (around phenyl ring)
DREIDING (Mayo, Olafson and Goddard, J. Phys. Chem. 94, 8897 (1990))
Torsion: checked to reproduce ab initio torsion potential for model system (G. Gao)
PAA1 (C18)
good
PAA2 (C18/C1)
good
The same MW
PAA3 (C22)
poor
PAS2 (C18/C1)
very poor
The same side chain distribution
n-heptane (n-C7)
n-C31 or n-C32
n-dotriacontane (n-C32)
(m.p.183 K; b.p. 372 K)
(amorphous; m.p.~340 K)
(crystalline)
n-C7 (liquid)
calc.(293 K)
expt’l
Density (g/cm3)
0.672  0.002
0.6838 (0.6795 at 298 K)
Hvap (kcal/mol)
8.87  0.3
8.74  0.004
solubility parameter
15.2 MPa1/2
15.1 MPa1/2
n-C32(amorphous)
calc.
expt’l
Density (g/cm )
0.816  0.003
0.8124
n-C32(crystal)
calc.
expt’l
Hvap (kcal/mol)
52.84  0.8
53.44  0.24 at 298 K
3
Calc.
• Average from 200-600 ps
NPT dynamics
• error from std. dev. of
block averages
Expt’l
• J. Chem. Eng. Data
9, 231 (1994)
• CRC handbook of
chemistry and physics
long alkanes in oil selectively partition toward the inhibitors
making them less available to nucleate a wax crystal
MD simulations started at various positions of n-C32 w.r.t. PAA1 in n-C7 bath
Unsequestered wax at 293 K
<PE> = -741  5* kcal/mol (100-200 ps)
*Error estimated by the standard deviation
between four 25-ps block average
Sequestered wax at 293 K
<PE> = -739  12* kcal/mol (100~200 ps)
No energy gain after sequestering
Close contact
Crystallization 1
CED = +318%
Very favorable
4. Crystalline pure n-C32
1. Amorphous pure n-C32
- Additive E << 0
Segregation CED = 55%
+ Additive
E << 0
Incorporation CED = -17%
Crystallization 2
CED = +80%
Less favorable
than above
2. Amorphous n-C32 with additive
3. Crystalline n-C32 with additive
(1  2  3  4) is slower than (1  4). (Crystallization is delayed with additive.)
before
PAA1
in n-C7
(E1)
pure
n-C31
(E2)
after
pure
n-C7
(E3)
PAA1
in n-C31
(E4)
E(incorporation) = Eafter-Ebefore = (E3+E4)-(E1+E2) = (E4-E2)-(E1-E3) = Eint(C31)-Eint(C7)
J/m2
Eint(n-C7)*
Eint(n-C31)*
Eint(n-C7-to n-C31)
PAA1
-68.5  0.8
-73.7  0.1
-5.2  0.8
PAA2
-68.9  0.3
-74.9  0.2
-6.0  0.4
PAA3
-67.6  0.2
-73.8  0.6
-6.2  0.6
PAS2
-69.8  0.9
-76.4  0.3
-6.6  0.9
Eincorporation per area (J/m2)
*Interaction energy between inhibitor with oil/wax
*averaged over 200~600 ps of MD simulations
*normalized by average contact area
*error estimated from duplicate runs for each system
-3
-4
-5
-6
-7
No correlation or
reverse correlation to expectation
-8
-9
0
20
40
60
80
Relative wax deposit (%)
100
120
Gauche population (%)
40
pure c31
c31/PAA2
c31/PAA3
c31/PAS2
35
30
25
20
pure C31(am)
15
C31/PAA2
10
C31/PAS2
5
pure C31(cr)
0
0
10
20
30
1
40
7
9 11 13 15 17 19 21 23 25 27
80
fluctuation*
Pure n-C31
21.6  7.3
2.52
n-C31 with PAA2
22.5  7.3
2.61
n-C31 with PAA3
23.3  6.7
3.04
n-C31 with PAS2
23.2  6.4
2.92
*average over 55 n-C31’s of standard deviation
of end-to-end distance along time 200-600 ps MD
Switch frequency (/ns)
average
5
torsion number (1 to n-4)
end-to-end distance (A)
End-to-end distance
3
70
pure C31(amor)
60
C31/PAA1
50
C31/PAS2
pure C31(cr)
40
30
20
Counted each 1ps
10
0
1
3
5
7
9 11 13 15 17 19 21 23 25 27
torsion number (1 to n-4)
Incorporated inhibitors disturb conformation relaxation of wax for crystallization? No
based on the difference in efficiency between hydrophilic PAA and hydrophobic PAS
based on the efficiency increase when inhibitor is added initially
Wax deposit
Preliminary study:
adsorption of inhibitor on a-Fe2O3, a model of pipewall
0.5
PAS2
PAA3
From 40~120 ps MD at solid(fix)-vacuum interface
PAA1
1.0
1.5
2.0
2
Adsorption energy per area (kcal/mol/A )
a-Fe2O3 force field
S. Jiang, et al. J. Phys. Chem. 100, 15760 (1996)
Sequestering mechanism? No.
No energy difference between sequestered and unsequestered state
There is no preference for wax molecules to be sequestered by inhibitor.
Incorporation-Perturbation mechanism? No.
It cannot explain the difference in efficiency between PAA and PAS.
Adsorption of inhibitor on hydrophilic surface (e.g. a-Fe2O3)
It looks good so far, but it needs more work.
Larry Smarr (U. Illinois) for supercomputer allocation at NCSA
Yanhua Zhou for a-Fe2O3 structure and force field