Simulation of enhanced oil recovery in naturally fractured reservoirs using dual-porosity models
Abstract
Carbonate reservoirs hold more than half of the world’s remaining oil reserves and around
40% of the world’s remaining gas reserves. Producing hydrocarbons from carbonate
reservoirs can be very challenging as carbonate reservoirs are often highly heterogeneous,
fractured, and with mixed- or oil-wet wettability, leading to low recovery factors after
primary and secondary recovery methods. Surfactant injection is a chemically enhanced
oil recovery method that can increase the recovery from fractured carbonate reservoirs by
reducing the interfacial tension between wetting and non-wetting phases and reducing the
capillary trapping forces. Surfactants can also change the wettability of oil- or mixed-wet
reservoir rocks to more water-wet, which promote spontaneous imbibition of the wetting
phase and increase the recovery.
This thesis aims to identify and analyse the limitations of dual-porosity models when
predicting the performance of chemically enhanced oil recovery methods in naturally
fractured reservoirs. Dual-porosity models are an idealised concept used to model
fractured porous media. Dual-porosity models consist of two distinctive domains: the
low-permeability matrix (providing storage) and the high-permeability fractures
(providing flow), which are connected via a transfer function that describes the fluid
exchange between both domains.
The central hypothesis of this thesis is that dual-porosity models do not capture all the
underlying physiochemical processes during chemically enhanced oil recovery, and
hence there is a need to develop improved solutions to model fracture-matrix transfer for
these applications more accurately. This thesis demonstrates how multiple interacting
continua (MINC) or vertical refinement (VR) models can be optimised without increasing
the complexity of the transfer functions itself to yield significantly more accurate
recovery results when modelling chemically enhanced oil recovery in fractured reservoirs
compared to the classical dual-porosity model or unoptimised MINC and VR models.
Furthermore, this thesis shows that accurately configured MINC and VR models are
important when performing large-scale reservoir simulations because inadequately
configured dual-porosity models add another uncertainty when predicting reservoir
performance during waterflooding or surfactant injection in fractured reservoirs.
As a key result, this thesis clearly demonstrates the potential of using optimised MINC
and VR models to analyse the performance of fractured reservoirs and highlight the error
associated when using unoptimised MINC and VR or classical dual-porosity models. The findings of this thesis can be applied in other scenarios, including contaminant transport
in groundwater resources, heat flow in high-enthalpy geothermal applications, reactive
transport during subsurface flow, or CO2 storage.