Numerical and machine learning modelling of the impact of hydrolysis on oil recovery by polymer flooding
File(s)
Author(s)
Khelifa, Mansour
Type
Thesis
Abstract
We investigate the impact of non-isothermal neutral hydrolysis on polymer flooding and determine which parameters have the greatest impact on the modelling of hydrolysis. Experimental
investigations using polyacrylamide polymers in solution have shown that neutral hydrolysis
can lead to a significant increase or decrease in the solution viscosity. This could change the performance of polymer flooding using these chemicals as the magnitude and nature of hy drolysis depends upon several factors including the water salinity, pH and temperature. There
is also a paucity of data in the literature for many of the parameters controlling the rate of
hydrolysis resulting in further uncertainty as to the importance of hydrolysis in reservoir ap plications. A fully implicit, two-phase, finite volume simulator was developed that models
the influence of salinity and temperature on the degree of polymer hydrolysis with polymer
viscosity and hence on the dynamics of a polymer flood. This model includes heat exchange
with the formation surrounding the reservoir. A degree of hydrolysis and salt concentration
dependent adsorption model was developed and implemented. The interactions between the
polymer ions and the counter ions occurring naturally in the connate water or injected wa ter lead to a non-monotonic variation of the polymer viscosity with the degree of hydrolysis
of the polymer. This was accounted for by the development of a novel correlation based on
Huggins empirical equation. A machine learning algorithm was developed and trained by the
data generated by the developed simulator to be able to identify the parameters that had
the highest impact on the polymer flood performance. The sampling of data in the space of
parameters was done using an experimental design strategy. A two-level factorial fractional
experimental design was implemented to determine which parameters had the most influence
on polymer flooding performance. Dimensionless analysis shows that the system of equations
depends upon 23 independent dimensionless scaling groups. Eight of these groups were evalu ated in the experimental design. A systematic literature review provided the range of values of
these dimensionless groups. The subset of groups was chosen following preliminary analysis of
the dimensionless groups. The analysis shows that neutral hydrolysis can increase or decrease
cumulative recovery by up to 4%. It can further modify slug injection by changing the degree
of viscous cross-flow or fingering of chase water into the polymer. This is due to the hydrolysis
zone lagging behind the polymer front. The main controls on recovery are the polymer, water
and oil viscosity ratios together with the salt-polymer interaction coefficient. In the reaction
model, the Damk¨ohler number and the Arrhenius numbers are most important in determining
the degree of hydrolysis at the polymer shock front and behind the shock front. The impact of
the degree of hydrolysis at the shock front and behind the front are highly dependent on the
salt concentration. This confirms that it is important to control the solution salt concentration
when planning injection of hydrolysed polyacrylamide.
The relationship between adsorption and the degree of hydrolysis was reviewed using data from
the literature and it was shown that the adsorption decreases with the degree of hydrolysis until
30% of hydrolysis. Above 30% hydrolysis, the amount of adsorbed polymer shows little variation
with a change of the degree of hydrolysis.
The study of the impact of adsorption on reservoir performances has shown that the impact of
this mechanism can be minimized in reservoirs that have a high temperature. Indeed, under
high temperatures, the reaction is extremely fast which allows a rapid increase of the degree of
hydrolysis and therefore a decrease of adsorption. This observation makes it possible to inject
polymers that have a very low degree of hydrolysis in reservoirs that have high temperatures
without significantly increasing the risk associated with adsorption. The injection of polymers
having a low degree of hydrolysis can potentially reduce the mechanical degradation of the
polymer near the injection region and improve the injectivity of the polymer. The impact of
hydrolysis on cross-flow and on oil recovery and on water-cut was assessed by simulating polymer
flooding through heterogeneous reservoirs. First, a two-layered reservoir was studied and it was
showed that hydrolysis can have an important impact on both cross-flow and on the recovery
factor when the initial oil-polymer viscosity ratio is close to one, when the permeability values
between the two layers are close to each other and when the initial degree of hydrolysis of the
injected polymer is low (i.e. less than 0.2). More complex permeability distributions were also
included in the investigations. We focused on inter-bedded reservoirs, upward or downward
fining reservoirs, and on normally distributed permeability fields. We also investigated the
impact of hydrolysis on SPE10 model2. It was shown that the hydrolysis has the highest
impact on the cross-flow and on the recovery factor in the inter-bedded reservoir. This was
explained by the fact that in these reservoirs, high pressure gradients exists due to the difference
in the permeability values between the adjacent layers which leads to a large amount of the
cross-flow and therefore to a better oil recovery. In the same time the cross-flow itself is affected
by the viscosity field that is highly dependent on the degree of hydrolysis.
investigations using polyacrylamide polymers in solution have shown that neutral hydrolysis
can lead to a significant increase or decrease in the solution viscosity. This could change the performance of polymer flooding using these chemicals as the magnitude and nature of hy drolysis depends upon several factors including the water salinity, pH and temperature. There
is also a paucity of data in the literature for many of the parameters controlling the rate of
hydrolysis resulting in further uncertainty as to the importance of hydrolysis in reservoir ap plications. A fully implicit, two-phase, finite volume simulator was developed that models
the influence of salinity and temperature on the degree of polymer hydrolysis with polymer
viscosity and hence on the dynamics of a polymer flood. This model includes heat exchange
with the formation surrounding the reservoir. A degree of hydrolysis and salt concentration
dependent adsorption model was developed and implemented. The interactions between the
polymer ions and the counter ions occurring naturally in the connate water or injected wa ter lead to a non-monotonic variation of the polymer viscosity with the degree of hydrolysis
of the polymer. This was accounted for by the development of a novel correlation based on
Huggins empirical equation. A machine learning algorithm was developed and trained by the
data generated by the developed simulator to be able to identify the parameters that had
the highest impact on the polymer flood performance. The sampling of data in the space of
parameters was done using an experimental design strategy. A two-level factorial fractional
experimental design was implemented to determine which parameters had the most influence
on polymer flooding performance. Dimensionless analysis shows that the system of equations
depends upon 23 independent dimensionless scaling groups. Eight of these groups were evalu ated in the experimental design. A systematic literature review provided the range of values of
these dimensionless groups. The subset of groups was chosen following preliminary analysis of
the dimensionless groups. The analysis shows that neutral hydrolysis can increase or decrease
cumulative recovery by up to 4%. It can further modify slug injection by changing the degree
of viscous cross-flow or fingering of chase water into the polymer. This is due to the hydrolysis
zone lagging behind the polymer front. The main controls on recovery are the polymer, water
and oil viscosity ratios together with the salt-polymer interaction coefficient. In the reaction
model, the Damk¨ohler number and the Arrhenius numbers are most important in determining
the degree of hydrolysis at the polymer shock front and behind the shock front. The impact of
the degree of hydrolysis at the shock front and behind the front are highly dependent on the
salt concentration. This confirms that it is important to control the solution salt concentration
when planning injection of hydrolysed polyacrylamide.
The relationship between adsorption and the degree of hydrolysis was reviewed using data from
the literature and it was shown that the adsorption decreases with the degree of hydrolysis until
30% of hydrolysis. Above 30% hydrolysis, the amount of adsorbed polymer shows little variation
with a change of the degree of hydrolysis.
The study of the impact of adsorption on reservoir performances has shown that the impact of
this mechanism can be minimized in reservoirs that have a high temperature. Indeed, under
high temperatures, the reaction is extremely fast which allows a rapid increase of the degree of
hydrolysis and therefore a decrease of adsorption. This observation makes it possible to inject
polymers that have a very low degree of hydrolysis in reservoirs that have high temperatures
without significantly increasing the risk associated with adsorption. The injection of polymers
having a low degree of hydrolysis can potentially reduce the mechanical degradation of the
polymer near the injection region and improve the injectivity of the polymer. The impact of
hydrolysis on cross-flow and on oil recovery and on water-cut was assessed by simulating polymer
flooding through heterogeneous reservoirs. First, a two-layered reservoir was studied and it was
showed that hydrolysis can have an important impact on both cross-flow and on the recovery
factor when the initial oil-polymer viscosity ratio is close to one, when the permeability values
between the two layers are close to each other and when the initial degree of hydrolysis of the
injected polymer is low (i.e. less than 0.2). More complex permeability distributions were also
included in the investigations. We focused on inter-bedded reservoirs, upward or downward
fining reservoirs, and on normally distributed permeability fields. We also investigated the
impact of hydrolysis on SPE10 model2. It was shown that the hydrolysis has the highest
impact on the cross-flow and on the recovery factor in the inter-bedded reservoir. This was
explained by the fact that in these reservoirs, high pressure gradients exists due to the difference
in the permeability values between the adjacent layers which leads to a large amount of the
cross-flow and therefore to a better oil recovery. In the same time the cross-flow itself is affected
by the viscosity field that is highly dependent on the degree of hydrolysis.
Version
Open Access
Date Issued
2020-03
Date Awarded
2021-09
Copyright Statement
Creative Commons Attribution NonCommercial NoDerivatives Licence
Advisor
Blunt, Martin
Sponsor
Equinor (Firm)
Publisher Department
Earth Science & Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)