You are on page 1of 9

Fuel 235 (2019) 1266–1274

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Experimental evaluation of RAFT-based Poly(N-isopropylacrylamide) T


(PNIPAM) kinetic hydrate inhibitors
Juwoon Parka, Hyunho Kima, Kelly Cristine da Silveirab,c, Qi Shengc,e, Almar Postmad,
⁎ ⁎
Colin D. Woodc, , Yutaek Seoa,
a
Department of Naval Architecture and Ocean Engineering, Research Institute of Marine Systems Engineering, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul
08826, Republic of Korea
b
Polytechnic Institute of Rio de Janeiro, Rio de Janeiro State University, Nova Friburgo, Brazil
c
CSIRO Australian Resources Research Centre, Kensington, WA 6152, Australia
d
CSIRO Manufacturing, Clayton, VIC 3168, Australia
e
Curtin University of Technology, Kensington, WA 6152, Australia

A R T I C LE I N FO A B S T R A C T

Keywords: As the oil and gas industry produces hydrocarbons from deeper waters and colder regions the issue of hydrate
Kinetic hydrate inhibitor formation becomes more serious. As a result, hydrate inhibition has focused on kinetic hydrate inhibitors (KHI)
PNIPAM and anti-agglomerants (AA) as an alternative to the existing approaches which involves injecting vast quantities
Polymer architecture of thermodynamic inhibitors. In this research, we evaluated the effect of different architectures (linear and
RAFT polymerization
branched) of poly(N-isopropylacrylamide) (PNIPAM) polymers synthesized using reversible addition−-
fragmentation chain-transfer (RAFT) polymerization. Unlike non-reversible deactivation radical polymerisation
(RDRP) synthetic routes this generates accurately controlled KHI candidates with target molecular weight,
narrow molecular weight distributions and controlled architecture, so that the effect on hydrate inhibition can
be more accurately assessed. The RAFT-based polymers (linear and branched) were compared to a commercially
available linear PNIPAM synthesized via non-RDRP radical polymerization and control groups (pure water, PVP,
and Luvicap). The hydrate experiments were performed in a high pressure autoclave with continuous cooling
under different cooling rates (0.25 K/min, 0.033 K/min, and 0.017 K/min). In addition, a cold restart was si-
mulated using constant subcooling. The results regarding subcooling temperature, onset time, and hydrate
fraction with resistance-to-flow were compared to known KHIs. These revealed that a linear PNIPAM-
MacroRAFT polymer delayed the hydrate nucleation with similar performance to known KHIs (eg., PVP and
Luvicap). However, a branched polymer showed the best performance in terms of hydrate fraction and re-
sistance-to-flow among all of the systems tested in this study. These data provide valuable information regarding
linear and branched PNIPAM-MacroRAFT polymers by demonstrating their ability to delay hydrate formation
but also in preventing hydrate agglomeration. These findings confirm that polymer architecture can effect hy-
drate inhibition.

1. Introduction methanol has been conventionally used for hydrate prevention in the
subsea flowlines. These inhibitors function by shifting the hydrate
Gas hydrates are nonstoichiometric crystalline compounds that are equilibrium conditions to a higher pressure and lower temperature re-
formed when water and light hydrocarbon molecules are present under gion. The amount of THIs that are injected is dictated by the hydrate
low temperature and high pressure conditions [1–3]. Recently, the equilibrium conditions and the amount of water in the flowlines
production of offshore oil and gas has moved to deeper and colder re- [1,4–7]. However, the search for new hydrocarbon sources has moved
gions which is more favorable for hydrate formation, increasing the risk the industry to more extreme environments that require higher THI
of hydrate blockage in flowlines. The most common method to prevent concentrations. This increases the CAPEX and OPEX [8] for new field
hydrate formation is chemical injection. The injection of thermo- developments.
dynamic inhibitors (THIs) such as monoethylene glycol (MEG) or As a result, the oil and gas industry has been investigating other


Corresponding authors.
E-mail addresses: colin.wood@csiro.au (C.D. Wood), yutaek.seo@snu.ac.kr (Y. Seo).

https://doi.org/10.1016/j.fuel.2018.08.036
Received 21 June 2018; Received in revised form 26 July 2018; Accepted 8 August 2018
Available online 05 September 2018
0016-2361/ © 2018 Elsevier Ltd. All rights reserved.
J. Park et al. Fuel 235 (2019) 1266–1274

options to replace THIs including hydrate avoidance and risk manage- showing a wide compatibility with functional monomers, it becomes an
ment [1,9–11]. Under the concept of risk management, several options obvious tool for developing finely tuned KHIs. Seo et al. synthesized
have been suggested such as under inhibition of THI [12–15], low do- PVCap using RAFT and free radical polymerization (FRP) and showed
sage hydrate inhibitors (LDHIs) [3,4,7,16] and a combination of these better inhibition performance with polymers of narrow molecular
options [17,18]. Using the under inhibition of THI less THI is required weight distributions [29]. Reyes et al. suggested that polymers con-
which reduces operational cost and complexity. However, the potential taining predominantly non-amide-based monomers such as iso-
of plugging with under inhibited systems increases because this relies propenyloxazoline (IPOx) could be synthesized using RAFT and the
on the concentration of THI which determines the properties of the resulting polymers have high cloud points and they showed inhibition
hydrate slurry. The use of LDHIs has become an alternative to ther- of hydrate nuclei rather than growth inhibition [38]. Overall molecular
modynamic inhibitors due to the low concentrations required, usually weight distribution is one of the factors that can influence the hydrate
0.5–3.0 wt%, which offers an economic alternative to hydrate inhibi- inhibition efficiency.
tion. LDHIs are classified as kinetic hydrate inhibitors (KHIs) and anti- In terms of polymer architecture there are limited investigations,
agglomerates (AAs) [3,4,7]. AAs allow hydrate to form but they prevent Del Villano et al. revealed that the tacticity of poly(N,N-dialkylacryla-
the cohesion between hydrate particles resulting in hydrate slurries mide) influences the KHI performance [30]. The polymer tacticity af-
that easily flow and are suitable for fluid transport. The surface of hy- fected its interaction with water molecules, inducing different water
drate particle becomes hydrophobic with the AAs causing the disper- perturbation. A syndiotactic polymer showed better performance than
sion of particles into the oil phase and the reduction of capillary in- other tacticities. Chua et al. [31] reported that different tacticities of
teractions. However, AAs are limited by the water cut (40–60%) PNIPAM with similar molecular weight changed the KHIs performance.
because the hydrate slurry becomes too viscous with high hydrate In this study, we investigated the effect of polymer architecture on
fraction in the liquid phase. KHIs are water-soluble polymers that ty- hydrate formation by comparing linear and branched polymers. In
pically contain homo- or copolymers of N-vinylcarprolactam (VCap), N- particular, linear-PNIPAM and branched-PNIPAM that were synthe-
isopropylacrylamide (NIPAM) or N-vinylpyrrolidone (VP). These sized using RAFT polymerization. The results are compared to PNIPAM
structures are composed of the polyethylene backbone connected to a synthesized with traditional radical polymerization, and commercial
pendant group. Groups containing hydrogen bonding sites such as KHIs (PVP and Luvicap). The results are presented in three sections: 1)
amide and imide can interact with the water molecules which defers the hydrate kinetics at a specific subcooling temperature and onset time, 2)
hydrate nucleation. Research on KHIs has been conducted via mod- hydrate growth rate to observe the different growth characteristics, and
ification or synthesis of well-known commercial polymers including 3) resistance to flow for the agglomeration of hydrate particles in the
PVCap [19–21], PVP [22,23], and PNIPAM [24–26]. However, these presence of KHIs.
polymers have limited performance as evidenced by the low subcooling
temperature which is the difference between fluid and hydrate forma-
2. Experimental section
tion temperatures. The maximum subcooling for KHIs is reported to be
approximately 14 K, above which the KHI may not work effectively [4].
2.1. Materials
This problem can be overcome by synergistic inhibition with THIs such
as methanol and glycol.
Deionized water and decane were supplied by OCI and Sigma-
To evaluate the performance of KHIs in delaying hydrate nucleation
Aldrich, respectively. Luvicap poly(N-vinylpyrrolidone) (PVP), com-
the onset time and subcooling temperature are usually investigated. −
mercial linear poly(N-isopropylacrylamide) (PNIPAM) (Mn 30 000)
The subcooling temperature and onset time indicate that hydrate for-
were used as control groups. Luvicap EG HM was supplied by BASF and
mation is delayed during the residence time of the fluids in the flow-
PVP [K-15, molecular weight (MW) = 9000 Da] was purchased from
lines (onset time) even when there is a significant driving force derived
Ashland Chemical Co. Those chemicals were used without further
from the subcooling temperature. However, limitations exist with using
purification. The synthetic natural gas used for hydrate formation (CH4:
onset time for assessing KHIs because hydrate formation is stochastic.
90 mol%, C2H6: 6 mol%, C3H8: 3 mol%, and n-C4H10: 1 mol%) was
Crystal growth inhibition (CGI) is an alternative method that has been
provided by Special Gas (Korea).
considered for the reliable and rapid evaluation of KHIs behavior.
The following materials were purchased from Sigma Aldrich and
Anderson et al. reported that the stochastic nature could be overcome
used for synthesis purpose: N-isopropylacrylamide (NIPAM, 97%),
by focusing on the hydrate growth using the subcooling temperature
azobisisobutyronitrile (AIBN, 98%), 1,4-dioxane (ACS reagent,
versus hydrate growth rate.[27] This was possible because KHIs func-
≥99.0%), n-heptane (LC-MS grade) and cyanomethyl dodecyl trithio-
tion by adsorbing onto the hydrate surface which affects the hydrate
carbonate (98% , HPLC grade), the inimer/transmer RAFT agent was
growth. The inhibition region which shows P_T is divided into 4 in-
synthesized as described in the literature [32].
cluding Complete Inhibition Region (CIR), Slow Growth Rate Region
(SGR), Rapid Growth Region (RGR) and Slow Dissociation Rate Region
(SDR). The growth rate during hydrate formation plays a significant 2.2. Polymerization of linear- and branched-PNIPAM via RAFT
role in distinguishing the different growth stages [28]. Daraboina et al.
reported the formation and decomposition of natural gas hydrates in NIPAM (1.012 g, 8.94 × 10−3 mol), RAFT agent cyanomethyl do-
the presence of KHIs using the high pressure calorimetry, the stirred decyl trithiocarbonate (0.0256 g, 9.06 × 10−5 mol), and AIBN (1.5 mg,
reactor, and the spectroscopic methods [39–42]. The crystal growth of 9.14 × 10−6 mol) were weighed in an ampule and dissolved in dioxane
natural gas hydrates was also observed to investigate the effect of the (2.945 mL) to an approximate molar ratio of [NIPAM]:[RAFT]:
KHI on the hydrate crystal grwoth. [AIBN] = 1000:10:1. The mixture was degassed with four free-
Recently, research has been conducted to develop more efficient ze−pump−thaw cycles, sealed, and then heated at 333.15 K in a
polymeric KHIs using chain growth polymerization techniques such as thermostated oil bath for 19 h 11 min. The final polymers were purified
reversible deactivation radical polymerization (RDRP), anionic or ca- by repeated precipitation from the monomer/solvent mixture into n-
tionic polymerizations.[4] Among these methods, reversible addition- heptane until full removal of small molecule impurities and dried under
fragmentation chain transfer (RAFT) has been investigated for synthe- vacuum until constant weight.
sizing polymeric KHIs with narrow polydispersity index (PDI) and The above procedure was adapted to synthesise the branched
controlled molecular weight. It is well known that the performance of macroRAFT, by replacing the commercial RAFT agent with the inimer/
KHIs is influenced by molecular weight [4], with molecular weight transmer RAFT agent. [32] The cost of these RAFT agents will con-
control being accessible through RDRP techniques, and with RAFT tribute to the overall economics of the process and would need to be

1267
J. Park et al. Fuel 235 (2019) 1266–1274

considered. However, this is a proof-of-concept study to demonstrate 1) The autoclave was filled with water and decane, then placed in the
the utility of RAFT polymerisation and the economics would need to be water bath. The gas phase was slowly flushed three times at 1 MPa
further investigated using more inexpensive agents that are emerging. with synthetic natural gas to remove residual air, which was then
Gel permeation chromatography (GPC) was performed on a Shimadzu pressurized by 12 MPa at 297.15 K.
system equipped with a CMB-20A controller system, an SIL-20A HT 2) The liquid phase was agitated at 600 rpm until pressure is stabilized
autosampler, an LC-20AT tandem pump system, a DGU-20A degasser at 297.15 K. Once confirming the pressure stabilized, continuous
unit, a CTO-20AC column oven, an RDI-10A refractive index detector, cooling was commenced to achieve target temperature 277.15 K
and 4 × Waters Styragel columns (HT2, HT3, HT4, and HT5, each with a cooling rate of 0.25 K/min. Hydrate formation was observed
300 mm × 7.8 mm2, providing an effective molar mass range of during or after the cooling process depending on the inhibition
100–4 × 106). N,N-Dimethylacetamide (DMAc) (containing 4.34 g L−1 performance of PVCap dissolved in water phase.
lithium bromide (LiBr)) was used as an eluent with a flow rate of 1 mL/ 3) The temperature was kept for 10 h at 277.15 K, then ramped up to
− −
min at 353.15 K. Number (Mn ) and weight average (Mw ) molar masses 301.15 K. Hydrate formation process was completed as seen in no
were evaluated using Shimadzu LC Solution software. The GPC columns more pressure reduction in gas phase. The temperature was main-
were calibrated with low dispersity poly(methymethacrylate) (PMMA) tained for 3 h to eliminate the memory effect.
standards (Polymer Laboratories) ranging from 1 110 to 2 136
000 g mol−1, and molar masses are reported as PMMA equivalents. A To simulate the cold-restart operation, the constant subcooling was
3rd-order polynomial was used to fit the log Mp vs. time calibration used in this work. The first procedure to stabilize the pressure was

curve. The number average molar masses (Mn ) of linear-PNIPAM and identical, but the next procedures were as follows:
branched-PNIPAM were found to be 17 600 kDa and 35 200 kDa, re-
spectively, with a molar mass dispersity index of 1.07 and 1.48. 2) Once confirming the stabilized pressure, the temperature was cooled
The molecular structure of polymers RAFT based PNIPAM is shown down to 277.15 K without agitation and maintained for 4.5 h.
in Fig. 1. Hydrate formation was sometimes observed during this period, as
the thermal driving force was large enough to initiate the hydrate
nucleation. However, the hydrate fraction in liquid phase was small
2.3. Experimental procedure for high pressure autoclave measurements due to limited mass transfer.
3) After 4.5 h of shut-in periods, the agitation was commenced at
To evaluate the hydrate inhibition performance of the newly syn- 600 rpm. Fast hydrate formation was observed from pressure re-
thesized polymers, hydrate formation characteristics including onset duction, which were studied to calculate the hydrate growth rate
time and growth rate were determined from experiments using a high and fraction in liquid phase.
pressure autoclave. The reactor was equipped with a magnetic stirrer
coupling and anchor type impeller was made by 316 SUS having an A total of five repeat experiments were performed for each system
inner volume of 360 mL. The total volume of 80 mL liquid phase with (Linear PNIPAM + decane + natural gas, PNIPAM-MacroRAFT (bran-
water cut 60% was loaded into the autoclave. Pure water, 0.5 wt% PVP ched) solution + decane + natural gas, and PNIPAM-MacroRAFT
solution, and 0.5 wt% Luvicap solution were selected as control groups (linear) solution + decane + natural gas) to obtain the hydrate for-
to compare the inhibition performance. The PNIPAM-MacroRAFT mation characteristics and the hydrate inhibition performance.
(branched and linear) as well as Linear PNIPAM (radical polymerization Furthermore, the experiment for the control groups (pure water + de-
PNIPAM) were evaluated at the concentration of 0.5 wt%. The auto- cane + natural gas, PVP solution + decane + natural gas, and Luvicap
clave was placed in a water bath connected to external water chiller solution + decane + natural gas) was also conducted five times. All
(Jeiotech RW2025G, Korea) to control the experimental temperature. experiments were conducted under isochoric condition.
The temperature of the liquid phase in the autoclave was measured by The constant subcooling method was also used to evaluate the
PT-100 Ω (accuracy: ± 0.15 K) and the pressure of the gas phase was performance of linear and branched PNIPAM-MacroRAFT. After pres-
detected using a transducer (WIKA, A-10) (accuracy: ± 0.01 MPa over surizing the autoclave with natural gas to 12 MPa and saturating the
0–20 MPa range). In addition, the torque was monitored using a torque liquid phase by mixing at 600 rpm, the mixing was stopped and the
sensor (TRD-50KC, ± 0.3%) to observe the change of resistance-to- temperature was decreased from 297.15 to 277.15 K within an hour
flow. Temperature, pressure, and torque were recorded through a data without mixing. The temperature was maintained 277.15 K for five
acquisition system. hours, then mixing at 600 rpm was commenced to simulate a cold re-
The experimental procedure to study the hydrate formation char- start operation. For each PNIPAM-MacroRAFT, the constant subcooling
acteristics with continuous cooling is as follows: experiment was repeated five times.

p m
HN O
S S
C12H25 C 4H 9
S S CN S S
n n n
HN O HN O HN O

Free radical Linear Branched


Fig. 1. Molecular structure of PNIPAM and RAFT based PNIPAM polymers. Free radical synthesized PNIPAM (left); RAFT synthesized linear (middle); and branched
(right).

1268
J. Park et al. Fuel 235 (2019) 1266–1274

3. Result and discussion for 0.033 K/min. Linear PNIPAM-MacroRAFT showed the highest ΔTsub
of 9.26 K for 0.017 K/min. For the inhibition of hydrate nucleation,
3.1. Effect of polymer architecture on hydrate formation kinetics linear PNIPAM-MacroRAFT and commercial KHIs showed better per-
formance under fast cooling rate, suggesting the linear structure may
Reversible Addition Fragmentation chain Transfer (RAFT) is a provide high energy barrier for hydrate nucleation. The rank of the
controlled radical polymerization technique that uses as a chain inhibition performance from the ΔTsub for each cooling rate is as fol-
transfer agent (RAFT agent) to control the growth of the molecular lows:
chains. This allows polymers with controlled molecular weight, narrow 0.25 K/min: PVP > linear PNIPAM-MacroRAFT ≈ Luvicap
molecular weight distributions, specific architectures and high end- > branched PNIPAM-MacroRAFT > linear PNIPAM
group fidelity to be generated. The RAFT agent remains at the end of 0.033 K/min: Luvicap > Linear PNIPAM-MacroRAFT ≈ Branched
the polymerization as an end-group. The generation of the linear PNIPAM-MacroRAFT ≈ PVP ≈ Linear PNIPAM
PNIPAM was via polymerizing NIPAM with a commercial RAFT agent, 0.017 K/min: Linear PNIPAM-MacroRAFT ≈ Branched PNIPAM-
whilst the dendritically branched PNIPAM was accessed using an in- MacroRAFT ≈ Linear PNIPAM > PVP > Luvicap
imer/transmer RAFT agent. Even though the branched PNIPAM-MacroRAFT shows a slightly
The subcooling temperature (ΔTsub) is the driving force for hydrate weaker KHI performance compared to the linear PNIPAM-MacroRAFT,
nucleation and has been used to evaluate the performance of KHIs. It is both PNIPAM-MacroRAFT materials have comparable kinetic inhibition
defined as the difference between the equilibrium and hydrate onset performance to commercial KHIs (Luvicap and PVP).
temperatures. The kinetics of hydrate formation are dependent on the
subcooling temperature and onset time. For the polymers under in- 3.2. Hydrate formation characteristics under continuous cooling
vestigation various cooling rates were used to determine the effect of
polymer architecture on the performance of the KHI. The hydrate onset Hydrate growth rate and hydrate volume fraction in the liquid
is determined as the point where the pressure decreases substantially phase were measured to study the hydrate formation with PNIPAM-
during the cooling process or at the target temperature. In this study, MacroRAFT (linear and branched), Linear PNIPAM and control groups
the cooling rate was varied from 0.25 K/min to 0.033 K/min and then (pure water, PVP and Luvicap solutions). Recent studies have revealed
0.017 K/min. With a slow cooling rate, the subcooling temperature that KHIs also affect the hydrate growth due to their adsorption onto
decreased and the onset time increased. Faster cooling rate indicates a the hydrate crystal surface. The gas consumption during the hydrate
larger driving force for hydrates to form during the cooling process at a formation was calculated for each system by measuring the difference
certain time, hence it results in a shorter onset time and higher sub- in pressure between the experimental pressure and estimated pressure
cooling temperature. Each experiment was repeated over five cycles without hydrate formation [33].
and the mean value and standard deviation for the subcooling tem-
Pexp Vcell
perature (ΔTsub) and onset time (tonset) were obtained (Table 1). ΔnH , t = ⎛ ⎜ ⎟
⎞ −⎛ Pcal Vcell ⎞
The ΔTsub of the linear PNIPAM-MacroRAFT under 0.25 K/min, ⎝ zRT ⎠t ⎝ zRT ⎠t
0.033 K/min, and 0.017 K/min was 11.64 K, 10.60 K, and 9.26 K. For
where, ΔnH , t is the mole of gas consumption for hydrate formation at a
the branched PNIPAM-MacroRAFT polymer the ΔTsub was 10.60 K,
certain time, Pexp is an experimental pressure, Vcell is gas volume, R is the
9.97 K, and 8.96 K under 0.25 K/min, 0.033 K/min, and 0.017 K/min,
gas constant, T is gas temperature, Pcal is calculated pressure with no
respectively. The ΔTsub of PNIPAM-MacroRAFT was higher than in pure
hydrate assumption, z is compression factor simulated by Cubic Plus
water (4.71 K, 3.30 K, and 3.25 K under 0.25 K/min, 0.033 K/min, and
Association EOS from Multiflash [34].
0.017 K/min) which indicates a kinetic inhibition for the PNIPAM. To
The hydrate volume fraction of the liquid phase is calculated by the
estimate the effect of the RAFT group on the kinetic hydrate inhibition,
following equation.
the kinetics of Linear PNIPAM also showed ΔTsub of 10.07 K, 10.44 K,
and 9.38 K under the studied cooling rates. Vhyd
Φhyd =
Commercial KHIs were also investigated including Luvicap and poly Vhyd + Vdecane + Vresid, water
(N-vinylpyrrolidone) (PVP) to compare their performance to the syn-
thesized polymers. While the ΔTsub of 0.5 wt% PVP solution varied from where Φhyd is a hydrate volume fraction, Vhyd is a volume of hydrate,
8.46 K to 12.52 K under cooling rates between 0.017 K/min and 0.25 K/ Vresid, water is a volume of residual water after water conversion to hy-
min, the ΔTsub of Luvicap was 11.58 K, 11.43 K, and 7.93 K under the drate. For calculation of the hydrate volume fraction, a hydration
same cooling rates. PVP showed the best performance with the sub- number of 6.5 was used which was obtained by cage occupancy of
cooling temperature of 12.52 K for 0.25 K/min and Luvicap was the best structure II hydrate [35]. The relative torque was also calculated from
the ratio of torque at specific moment (τexp, t ) to torque when there was
Table 1 no hydrate (τnohyd, to ), suggesting how resistance-to-flow increased.
Hydrate kinetic characteristics of 0.5 wt% PNIPAM solutions (Linear and τexp, t
Branched PNIPAM-MacroRAFT, and Linear PNIPAM.) under various cooling τrel =
τnohyd, to
rates (0.25 K/min, 0.033 K/min, and 0.017 K/min). The value in parenthesis is
the standard deviation. Data for pure water, 0.5 wt% PVP, and 0.5 wt% Luvicap Table 2 presents initial growth rate, hydrate volume fraction at
solutions were presented in our previous work [26].
100 min after the onset, and the amount of water converted into hy-
Cooling rate ΔTsub (K) tonset (min) drate while varying the cooling rate. Fig. 3 shows the hydrate volume
(K/min) fraction as a function of time until 100 min after the nucleation as the
Linear PNIPAM-macroRAFT 0.25 11.64 (0.19) 77.33 (3.75)
initial growth rate is important to evaluate the performance of the KHIs.
0.033 10.4 (0.89) 358.13 (12.65) The time zero indicates the moment of hydrate nucleation. The hydrate
0.017 9.26 (0.14) 566.43 (11.43) fraction of pure water recorded the highest conversion from water to
Branched PNIPAM-macroRAFT 0.25 10.60 (0.09) 67.33 (2.96) hydrate under three different cooling rates. The initial growth rate for
0.033 9.97 (0.25) 309.36 (7.68)
pure water was 8.9 × 10−3 vol.frac./min and the hydrate fraction
0.017 8.96 (1.42) 532.77 (79.19)
Linear PNIPAM 0.25 10.07 (0.09) 48.83 (1.65) reached 0.43 at 0.25 K/min cooling rate. With the addition of Luvicap,
0.033 10.44 (0.60) 323.86 (20.54) the growth rate was reduced to 4.9 × 10−3 vol.frac./min while
0.017 9.38 (0.47) 576.63 (30.95) achieving the fraction 0.13 under 0.25 K/min cooling rate. The addition
of PVP also showed that the hydrate volume fraction and the growth

1269
J. Park et al. Fuel 235 (2019) 1266–1274

Table 2
Hydrate formation characteristics of 0.5 wt% PNIPAM solutions (linear
PNIPAM-MacroRAFT, branched PNIPAM-MacroRAFT, and, linear PNIPAM)
under various cooling rates (0.25 K/min, 0.033 K/min, and 0.017 K/min).
Standard deviation values in parenthesis.
Cooling rate rini (min−1) ϕhyd,100min ϕhyd,600min
(K/min)

Linear PNIPAM-MacroRAFT 0.25 9.0 × 10−3 0.28 (0.05) 0.39 (0.05)


0.033 8.2 × 10−3 0.18 (0.08) 0.36 (0.05)
0.017 1.6 × 10−3 0.22 (0.02) 0.39 (0.05)
Branched PNIPAM-MacroRAFT 0.25 1.9 × 10−3 0.16 (0.07) 0.36 (0.06)
0.033 2.0 × 10−3 0.16 (0.02) 0.38 (0.07)
0.017 0.4 × 10−3 0.03 (0.02) 0.28 (0.04)
Linear PNIPAM 0.25 3.8 × 10−3 0.13 (0.04) 0.26 (0.02)
0.033 5.8 × 10−3 0.09 (0.01) 0.22 (0.02)
0.017 1.1 × 10−3 0.06 (0.01) 0.22 (0.02)

rate were decreased to 0.29 and 6.4 × 10−3 vol.frac./min, respectively.


These results indicated that Luvicap showed better performance than
PVP in term of slow growth rate and less hydrate fraction in the liquid
phase [26]. For the cooling rate of 0.033 K/min and 0.017 K/min, Lu-
vicap also outperformed PVP.
Although linear PNIPAM-MacroRAFT showed the best performance
for delaying the hydrate onset as shown in Fig. 2, it did not effectively
suppress the growth of hydrate crystals. The initial growth rate for
linear PNIPAM-MacroRAFT was 9.0 × 10−3 vol.frac./min under
0.25 K/min cooling rate, which was close to that of pure water and was
higher than that of both Luvicap and PVP. The hydrate fraction was also
high enough to demonstrate that it was not effective at inhibiting hy-
drate growth. When the cooling rate was 0.017 K/min, the growth rate
was 1.6 × 10−3 vol.frac./min while the hydrate fraction was still 0.22.
However, the branched PNIPAM-MacroRAFT outperformed both Lu-
vicap and PVP. The initial growth rate was 1.9 × 10−3 vol.frac./min
and 2.0 × 10−3 vol.frac./min under 0.25 K/min and 0.033 K/min
cooling rate, respectively. The hydrate fraction reached approximately
0.15 for both cooling rates. For the 0.017 K/min cooling rate, the initial
growth rate was 0.4 × 10−3 vol.frac./min with hydrate fraction of 0.03
at 100 min. The growth curve for Linear PNIPAM was placed between
the linear and branched PNIPAM-MacroRAFT in early stage of hydrate
formation. For cooling rates of both 0.25 and 0.033 K/min, an inflection
point was observed for branched PNIPAM-MacroRAFT in Fig. 3 when
the hydrate fraction reached 0.03, after which the growth rate picked
up, resulting in a higher hydrate fraction than Linear PNIPAM.
Fig. 3 shows hydrate growth curves 3 for each KHI and demon-
strates that the structure of the polymer had a significant impact on the
hydrate inhibition performance. Our previous work suggested that
torque could be used to estimate the flowability of a hydrocarbon fluid
in the presence of hydrate particles [26]. When the hydrate fraction
increased to more than 0.10 in the liquid phase an increase in torque is
induced for pure water due to the agglomeration of hydrate particles. Fig. 2. Subcooling temperature of 0.5 wt% polymer groups (Linear PNIPAM,
Luvicap and PVP showed relatively stable torque during the hydrate linear PNIPAM-MacroRAFT and branched PNIPAM-MacroRAFT and control
formation under 0.25 K/min cooling rate, however, the torque fluc- groups (pure water, PVP, and Luvicap solutions) under various cooling rates (a)
tuated under 0.033 and 0.017 K/min, which may indicate the hydrate 0.25 K/min, (b) 0.033 K/min, and (c) 0.017 K/min.
particle were separated rather than a homogeneous dispersion in the
liquid phase. The torque changes as a function of hydrate fraction in the agglomeration of hydrate particles. The torque remained stable at
liquid phase for newly RAFT-synthesized PNIPAM are shown in Fig. 4. 0.017 K/min cooling rate in the presence of all synthesized PNIPAM
The branched PNIPAM-MacroRAFT showed slow torque increase polymers. Overall, the branched PNIPAM-MacroRAFT suppressed the
until the hydrate fraction reached 0.10 for 0.25 and 0.033 K/min, then torque during the hydrate formation. The linear PNIPAM-MacroRAFT
the torque either decreased steadily or rapidly returned to the baseline. showed limited performance in terms of controlling torque increase
However, the linear PNIPAM-MacroRAFT showed a rapid increase in when the subcooling temperature was varied. These results suggest that
torque during the early stages of hydrate formation before decreasing to the agglomeration of hydrate particles might be controlled in the pre-
the baseline with hydrate fraction 0.10 for 0.25 K/min cooling rate. The sence of effectively designed KHI molecules. The branched polymer
rapid torque increase was repeated for 0.033 K/min cooling rate when structure induces slow hydrate growth and can inhibit the agglomera-
the hydrate fraction was higher than 0.15. The Linear PNIPAM showed tion of hydrate particles as show in Figs. 3 and 4. KHIs synthesized
similar performance to those of branched PNIPAM-MacroRAFT, using conventional free radical polymerization have promising perfor-
suggesting that the branched functional groups can inhibit the mance but further improvements in the performance can be achieved

1270
J. Park et al. Fuel 235 (2019) 1266–1274

Fig. 4. Relationship between hydrate volume fraction and torque to observe the
resistance to flow during hydrate formation under (a) 0.25, (b) 0.033, and (c)
Fig. 3. Average hydrate volume fraction of 0.5 wt% polymer (Linear PNIPAM, 0.017 K/min cooling rate. (Red) Linear PNIPAM, (Blue) Linear PNIPAM-
Linear PNIPAM-MacroRAFT and Branched PNIPAM-MacroRAFT) and 0.5 wt% MacroRAFT, (Green) Branched PNIPAM-MacroRAFT. (For interpretation of the
control groups (pure water, PVP, and Luvicap) under various cooling rates (a) references to colour in this figure legend, the reader is referred to the web
0.25 K/min, (b) 0.033 K/min, and (c) 0.017 K/min. Time 0 means the onset version of this article.)
moment [26].

growth rate, and are able to inhibit the agglomeration of hydrate par-
by controlling the structure of the KHI using RAFT polymerization. The
ticles if the structure of the polymer is controlled. However, industry
difference in hydrate inhibition performance between linear and
reports [3,4] have suggested that their performance might be limited
branched PNIPAM-MacroRAFT clearly suggests that the design of KHI
under high subcooling conditions i.e. more than 14 K, especially if the
structures is vital to achieve the desired inhibition performance. This is
hydrocarbon fluid resides inside the hydrate formation region for pro-
in line with our previous work [25,26] which also demonstrated that
longed duration. Aman et al. observed that a hydrate film was formed
controlling the structure of KHIs can modulate the peformance, which is
without mixing and this played a role of hydrate nucleation for fast
not easily accessible using conventional free radical techniques.
hydrate growth during a cold restart operation [36]. Our previous work
suggested that Luvicap may be effective at interacting with hydrate
3.3. Hydrate formation characteristics in cold restart operation nuclei at 1.0 wt% to suppress their nucleation and growth in the early
stages, at a low mixing rate of 200 rpm. But loosing its efficacy when
As presented above, KHIs can delay hydrate nucleation, reduce its the mixing rate increased to 600 rpm [37], where the onset time was

1271
J. Park et al. Fuel 235 (2019) 1266–1274

Table 3
Hydrate growth rate of mixing at start moment and hydrate volume fraction at
100 min and final moment.
rmax rlinear ϕhyd, ϕhyd,final τmax
(min−1) (min−1) 100min

Linear PNIPAM-MacroRAFT 7.6 × 10−2 4.3 × 10−2 0.19 0.22 3.5


Branched PNIPAM-MacroRAFT 7.2 × 10−2 3.0 × 10−2 0.17 0.21 1.5

dropped to 4.2 min. In this work, we performed hydrate formation


experiments to investigate the performance of both linear and branched
PNIPAM-MacroRAFT under a simulated cold restart operation with
high subcooling condition of 16 K. The concentration of PNIPAM was
maintained at 0.5 wt%, and the PNIPAM solution was kept at 277.15 K
and 10.5 MPa (subcooling temperature of 16 K) for approximately five
hours before commencing the mixing at 600 rpm.
Table 3 presents the hydrate fraction, maximum growth rate upon
mixing, linear growth rate before reaching an inflection point, hydrate
fractions at 100 min and at the end of experiments, and maximum re-
lative torque observed during the experiment. Fig. 5 (a) and (b) shows
the hydrate fraction along with the torque changes upon mixing as a
function of time for linear PNIPAM-MacroRAFT and branched PNIPAM-
MacroRAFT, respectively. It was observed that a thin hydrate film
formed on the surface of PNIPAM solutions during the period when the
system was not mixed. Upon commencing stirring (600 rpm), the hy-
drate film broke up into small pieces and dispersed into the liquid phase
due to turbulent mixing, this was followed by rapid hydrate formation
as shown in Fig. 5. For both linear and branched PNIPAM-MacroRAFT,
the initial growth rates were 7.6 × 10−2 and 7.2 × 10−2, respectively,
and the rates were about eight times larger than that of the linear
PNIPAM-MacroRAFT, in the continuous cooling experiment,
9.0 × 10−3 min−1. This result is in agreement with our previous work
and of those from Aman et al., observing fast hydrate growth initiated
from the previously-formed hydrate pieces. [36] Both the linear and
branched PNIPAM-MacroRAFT showed limited inhibition performance
on hydrate nucleation when the solution was maintained at the high
subcooling condition of 16 K for five hours.
However, once the hydrates started to grow the branched PNIPAM-
MacroRAFT showed better performance on hydrate growth inhibition
than the corresponding linear PNIPAM-MacroRAFT. The initial growth
rate quickly reduced from 7.6 × 10−2 to 4.3 × 10−2 for linear
PNIPAM-MacroRAFT before reaching an inflection point at 18 min. The Fig. 5. Relationship between hydrate growth and relative torque during a si-
mulated cold restart. Time zero indicates start of mixing. (a) Linear PNIPAM-
hydrate fraction was about 0.17 and the relative torque showed a
MacroRAFT; (b) Branched PNIPAM-MacroRAFT. (c) Resistance to flow with
maximum of 3.5 at the inflection point, indicating the highest re-
increasing hydrate fraction.
sistance-to-flow. After the inflection point, the hydrate growth rate
reduced and the relative torque returned to the original value. For the
branched PNIPAM-MacroRAFT, the growth rate was reduced to the autoclave wall by the impeller. After that the contact area between
3.0 × 10−2 after the initial fast growth and the hydrate fraction at the the slurry and the impeller was reduced. In the case of the branched
inflection point was at 0.10. The torque increased to 1.5 around the PNIPAM-MacroRAFT (Fig. 6 (b)) the hydrate particles appear to be
inflection point, but decreased slowly while increasing hydrate fraction. brittle and well crushed into small pieces by the impeller. The second
These results suggest that the hydrate formation characteristics were image shows the viewing window covered with hydrate particles but
different in linear and branched PNIPAM-MacroRAFT solutions. Fig. 5 not a slurry, thus these hydrate particles did not induce an increase of
(c) depicts the torque changes as a function of hydrate fraction in the liquid phase viscosity. These images clearly show the different hydrate
liquid phase. Linear PNIPAM-MacroRAFT showed increasing relative properties in the presence of linear and branched PNIPAM-MacroRAFT.
torque with increasing hydrate fraction, but branched PNIPAM-Mac- Further investigations will be carried out, however, these results sug-
roRAFT showed little increase in relative torque. gested that branched polymer structure may better interact with
Fig. 6 shows images captured during the hydrate formation. The growing hydrate crystals than a linear analogue. More studies will also
first images were taken when the mixing was initiated, with the second be carried out to quantitatively analyze the effect of polymer structure
images taken 20 min after. The final images were taken at the end of the on the growth and agglomeration of hydrate particles.
experiment. For linear PNIPAM-MacroRAFT, Fig. 6 (a), growing hy-
drate particles in the liquid phase induced a viscosity increase due to 4. Conclusions
interaction between hydrate particles and liquid phase. As seen in the
image at 20 min, which is just after the inflection point in growth curve, This work investigated the kinetic hydrate inhibition performance
the hydrate and liquid mixture looked like a slurry being pushed toward of RAFT synthesized PNIPAM, using hydrate onset conditions, growth

1272
J. Park et al. Fuel 235 (2019) 1266–1274

(a)

[t = 0 min] [t= 20 min] [t = 300 min]


(b)

[t = 0 min] [t = 20 min] [t = 300 min]


Fig. 6. Images of hydrate phase during hydrate formation. (a) Linear PNIPAM-MacroRAFT; (b) Branched PNIPAM-MacroRAFT.

rate, hydrate fractions, and torque changes while varying the cooling Press/Taylor & Francis; 2008.
rate. Both linear and branched PNIPAM-MacroRAFT show the sub- [2] Sloan ED. Fundamental principles and applications of natural gas hydrates. Nature
2003;426:353–63.
cooling behavior similar to those of commercial KHIs, Luvicap and PVP. [3] Kelland MA. Production chemicals for the oil and gas industry. 2nd ed. Boca Raton,
Once the hydrate forms, a branched PNIPAM-MacroRAFT was effective FL: CRC Press; 2014.
in reducing the growth rate and maintaining well-dispersed hydrate [4] Kelland MA. History of the development of low dosage hydrate inhibitors. Energy
Fuels 2006;20:825–47.
particles in the liquid phase. When decreasing the cooling rate the [5] Ballard A. Flow assurance lessons: the mica tieback. In: Offshore technology con-
subcooling temperature slightly changed but hydrate growth rate de- ference, Houston, TX, May 1–4; 2006.
creased significantly, suggesting the importance of the thermal driving [6] Lee J, Hampton B, Alapati RR, Sanford EA, O’Brien S. Innovative technique for
flowline plug remediation. In: Offshore technology conference, Houston, TX, May
force to form the hydrate crystals. A cold restart was simulated using a 4–7; 2009.
high subcooling (16 K) for both linear and branched PNIPAM- [7] Sloan ED, Koh CA, Sum AK. Natural gas hydrates in flow assurance. Amsterdam:
MacroRAFT which showed a significant impact on the hydrate growth Elsevier; 2010.
[8] Sanford E, Alapati RA. A new, field-proven, cost-effective solution for MEG re-
rate and torque changes as the branched PNIPAM-MacroRAFT was ef-
generation unit issues in offshore Australia gas production. APPEA J.
fective in interacting with growing hydrate particles. Research on KHIs 2011;51:193–200.
have primarily been focusing on their performance to delay the hydrate [9] Sloan ED, Koh CA, Sum AK, Ballard AL, Shoup GJ, McMullen N, et al. Hydrates:
onset. However, our results suggested that KHIs would play an im- state of the art inside and outside flow lines. J Pet Technol 2009;61:89–94.
[10] Gao S. Hydrate risk management at high water-cuts with anti-agglomerant hydrate
portant role in controlling the hydrate growth rate and the interaction inhibitors. Energy Fuels 2009;23:2118–21.
between hydrate particles. [11] Tohidi B, Anderson B, Chapoy A, Yang J, Burgass RW. Do we have new solutions to
the old problem of gas hydrates? Energy Fuels 2012;26:4053–8.
[12] Di Lorenzo M, Aman ZM, Kozielski K, Norris BWE, Johns ML, May EF.
Acknowledgment Underinhibited hydrate formation and transport investigated using a single-pass gas
dominant flowloop. Energy Fuels 2014;28:7274–84.
This work was partially supported by CNPq, Brazil through PDJ [13] Johns ML, May EF. Formation of gas hydrate blockages in under-inhibited condi-
tions. Edinburgh, Scotland: United Kingdom; 2011.
grant to Kelly Cristine da Silveira (Proc. 150710/2017-8) and also [14] Abay H, Svartaas TM. Effect of ultralow concentration of methanol on methane
supported by the Technology Innovation Program (10060099) funded hydrate formation. Energy Fuels 2010;24:752–7.
by the Ministry of Trade, Industry & Energy (MI, Korea). [15] Yousif MH. Effect of underinhibition with methanol and ethylene glycol on the
hydrate-control process. SPE Prod. Facil. 1998;13:184–9.
[16] Clark LW, Anderson J. Low dosage hydrate inhibitors (LDHI): Further advances and
References developments in flow assurance technology and applications concerning oil and gas
production systems. In: international petroleum technology conference, Dubai, U.A.
E.; December 4−6, 2007.
[1] Sloan ED, Koh CA. Clathrate hydrates of natural gases. 3rd ed. Boca Raton, FL: CRC

1273
J. Park et al. Fuel 235 (2019) 1266–1274

[17] Zhao X, Qiu Z, Zhou G, Huang W. Synergism of thermodynamic hydrate inhibitors Fuels 2017;31:6358–63.
on the performance of poly (vinyl pyrrolidone) in deepwater drilling fluid. J. Nat. [30] Del Villano L, Kelland MA, Miyake GM, Chen EY-X. Effect of Polymer Tacticity on
Gas. Sci. Eng. 2015;23:47–54. the Performance of Poly(N, N-dialkylacrylamide)s as Kinetic Hydrate Inhibitors.
[18] Kim J, Shin K, Seo Y, Cho SJ, Lee JD. Synergistic Hydrate Inhibition of Energy Fuels 2010;24:2554–62.
Monoethylene Glycol with Poly(vinylcaprolactam) in thermodynamically under [31] Chua PC, Kelland MA, Hirano T, Yamamoto H. Kinetic hydrate inhibition of poly(N-
inhibited system. J Phys Chem B 2014;118:9065–75. isopropylacrylamide)s with different tactilities. Energy Fuels 2012;26:4961–7.
[19] Mady MF, Kelland MA. N, N-dimethylhydrazidoacrylamides. Part 2: High-cloud- [32] Zhang Y, Teo BM, Postma A, Ercole F, Ogaki R, Zhu M, et al. Highly-branched poly
point kinetic hydrate inhibitor copolymers with n-vinylcaprolactam and effect of (N-isopropylacrylamide) as a component in poly(dopamine) films. J Phys Chem B
pH on performance. Energy Fuels 2015;29:678–85. 2013;117(36):10504–5012.
[20] O’Reilly R, Ieong NS, Chua PC, Kelland MA. Crystal growth inhibition of tetra- [33] Cha M, Shin K, Seo Y, Shin JY, Kang S-P. Catastrophic growth of gas hydrates in the
hydrofuran hydrate with poly(N-vinyl piperidone) and other poly(N-vinyl lactam) presence of kinetic hydrate inhibitors. J Phys Chem A 2013;117:13988–95.
homopolymers. Chem Eng Sci 2011;66:6555–60. [34] Multiflash, version 4.4; KBC Advanced Technologies plc: Surrey, U.K.; 2014.
[21] Lucas EF, Spinelli LS, Khalil CN. Polymers applications in petroleum production. [35] Seo Y, Kang S-P, Jang W. Structure and composition analysis of natural gas hy-
Hoboken, NJ: Encyclopedia of Polymer Science and Technology; John Wiley & Sons drates: 13C NMR spectroscopic and gas uptake measurements of mixed gas hydrates.
Inc; 2002. J Phys Chem A 2009;113:9641–9.
[22] Daraboina N, Linga P. Experimental investigation of the effect of Poly-N-vinyl [36] Aman ZM, Akhfash M, Johns ML, May EF. Methane hydrate bed formation in a
pyrrolidone (PVP) on methane/propane clathrates using a new contact mode. Chem visual autoclave: cold restart and Reynolds number dependence. J Chem Eng Data
Eng Sci 2013;93:387–94. 2015;60:409–17.
[23] Qin HB, Sun ZF, Wang XQ, Yang JL, Sun CY, Liu B, et al. Synthesis and evaluation of [37] Sohn YH, Seo Y. Effect of monoethylene glycol and kinetic hydrate inhibitor on
two new kinetic hydrate inhibitors. Energy Fuels 2015;29:7135–41. hydrate blockage formation during cold restart operation. Chem Eng Sci
[24] Perrin A, Musa OM, Steed JW. The chemistry of low dosage clathrate hydrate in- 2017;168:444–55.
hibitors. Chem Soc Rev 2013;42:1996–2015. [38] Reyes FT, Malins EL, Becer CR, Kelland MA. Non-Amide Kinetic Hydrate Inhibitors:
[25] da Silveira KC, Sheng Q, Tian W, Fong C, Maeda N, Lucas EF, et al. High throughput Performance of a Series of Polymers of Isopropenyloxazoline on Structure II Gas
synthesis and characterization of PNIPAM-based kinetic hydrate inhibitors. Fuel Hydrates. Energy Fuels 2013;27:3154–60.
2017;188:522–9. [39] Daraboina N, Ripmeester J, Walker VK, Englezos P. Natural gas hydrate formation
[26] Park J, da Silveira KC, Sheng Q, Wood CD, Seo Y. Performance of Ppoly(N-iso- and decomposition in the presence of kinetic inhibitors. 1. High pressure calori-
propylacrylamide)-based kinetic hydrate inhibitors for nucleation and growth of metry. Energy Fuels 2011;25:4392–7.
natural gas hydrates. Energy Fuels 2017;31:2697–704. [40] Daraboina N, Linga P, Ripmeester J, Walker VK, Englezos P. Natural gas hydrate
[27] Anderson R, Mozaffar H, Tohidi B. Development of a crystal growth inhibition formation and decomposition in the presence of kinetic inhibitors. 2. Stirred reactor
based method for the evaluation of kinetic hydrate inhibitors. In: 7th International experiments. Energy Fuels 2011;25:4384–91.
Conference on Gas Hydrates; Edinburgh, U.K.; July 17–21, 2011. [41] Daraboina N, Ripmeester J, Walker VK, Englezos P. Natural gas hydrate formation
[28] Mozaffar H, Anderson R, Tohidi B. Reliable and repeatable evaluation of kinetic and decomposition in the presence of kinetic inhibitors. 3. Structural and compo-
hydrate inhibitors using a method based on crystal growth inhibition. Energy Fuels sitional changes. Energy Fuels 2011;25:4398–404.
2016;30:10055–63. [42] Kumar R, Lee JD, Song M, Englezos P. Kinetic inhibitor effects on methane/propane
[29] Seo SD, Paik H-J, Lim D-H, Lee JD. Effects of Poly(N-vinylcaprolactam) Molecular clathrate hydrate-crystal growth at the gas/water and water/n-heptane interfaces. J
Weight and Molecular Weight Distribution on Methane Hydrate Formation. Energy Cryst Growth 2008;310:1154–66.

1274

You might also like