Polymer
solutions are designed to develop a favorable mobility ratio between
the injected polymer solution and the oil–water bank being
displaced by the polymer. Subsequently, a more uniform volumetric
sweep of the reservoir is produced. Chemical and mechanical degradation
of the polymer solutions, on the other hand, reduce their viscosity
which significantly affects their performance. The primary objective
of this study is to investigate the effect of surface modification
of silica nanoparticles (NPs) on the effective viscosity of partially
hydrolyzed polyacrylamide (HPAM) and xanthan gum (XG) solutions at
different NP concentrations and temperatures. The chemical functionalization
of SiO2 NPs with carboxylic acids and silanes was confirmed
by FTIR measurements. The experimental results showed that the addition
of SiO2 NPs increased the viscosity of XG solutions due
to the formation of three-dimensional structures between the silica
NPs and the polymeric chains. The thickening effect of HPAM was improved
by the addition of silica NPs modified with 3-(methacryloyloxy)propyl]
trimethoxysilane (MPS), octyl triethoxysilane (OTES), and oleic acid-method
A (OAA). In addition, the HPAM and XG nanopolymer sols of modified
silica NPs showed more temperature and brine tolerance than that of
unmodified silica NPs. A model was developed based on multilayer perceptron
(MLP) neural network for predicting viscosity of nanopolymer sols
using 9900 data points. The MLP model was trained by Bayesian Regularization
(BR), Levenberg–Marquardt (LM), Resilient Backpropagation (RB),
and Scaled Conjugate Gradient (SCG) algorithms. The results revealed
that the BR-MLP model outperformed the three other models and could
predict all the viscosity data with an average absolute relative error
of 2.46% and R
2 of 0.999.