Computational modeling of multiphase fluid flow in highly heterogenous problems with complicated geometries is a challenging problem for reservoir engineers, with a rich research in establishing best methods and approaches. The novelty in this work is centered around the implementation and comparison of simulation results from two software - the open source ICFESRT and the commercial software ECLIPSE - for a two-phase multiphase problem (oilwater) in both simple and complex geometries. The work involves: (a) implementation and
comparison of simulation results from the two software on three different, hypothetical but typical geometries; (b) consideration of a real field case and the associated data analysis, rock characterization, and geostatistics of a real field representative of a highly heterogeneous reservoir; and (c) implementation of both software on the real field case for predictions of oil production at the site, and comparison of the simulation results from the two software.
The initial comparison of simulation results for was carried out using three hypothetical (but common) geometries, these being: (a) a quarter five spot with one geological layer; (b) the same geometry as in (a) but with a vertical heterogeneity i.e. 5 different geological layers; (c) and lastly a full 5 spot with 5 different geological layers was implemented. Three different mesh resolutions were applied in both software and comparisons were carried out for mesh-independency. The results showed that in all these three scenarios, good agreement was observed between IC-FERST (coarse mesh) and ECLIPSE (fine mesh) with an average percentage difference at the production well ranging between 2.5% and 10.5% for the oil production and 12% and 26% for the water production.
Both the ICFERST and ECLIPSE were subsequently implemented on a real, heterogeneous field – which consisted of 25 producing wells and 8 injections wells. Prior to the software implementation, a data analysis and rock characterization was carried out –Using data from the 33 wells. The logging and core data (a total of 30,000 log readings and 1150 core samples) were utilized and a novel rock characterization technique -Balaha Rock Characterization Code- was implemented to allow for the optimal clustering of rock types within the reservoir, The rock characterization resulted in identifying 7 rock types with their unique porosity-hydraulic permeability relationships. Subsequently, geostatistical methods were implemented – which enabled populating the computational cells of the two software with the corresponding reservoir properties (porosity, hydraulic permeability). To achieve the property population into the
unstructured computational domain of the ICFERST software, a newly-developed script was written in Matlab and Python. The rock properties data populated on IC-FERST consist of porosity, permeability, relative permeability, capillary pressure and connate water saturation.
A further comparison between the IC-FERST simulation results with the corresponding ECLIPSE simulations was carried out – were all simulations were carried out for a period of 40 years. The percentage differences between the two software simulations were estimated for : (i) ten individual
production wells and (ii) the total of all production wells. The results showed that a good agreement exists between the IC-FERST and ECLIPSE simulations, with an average percentage difference for the total oil production of 10.5%, the total water production of 26% and the total water injection
of 14%. The results for the ten individual wells showed an average percentage difference of 15.5% ranging from 3 to 29% for the oil production in the late time period. Slightly higher differences were observed when the overall period was considered, due to the large difference at the early time
period of the simulation.
The results indicated that IC-FERST, when incorporating the necessary rock characterization information – which highlight the heterogeneity of the reservoir – can produce results that can compete with the industry standard ECLIPSE. Additional aspects need to be considered within the current real field IC-FERST simulation, the inclusion of possible fractures and faults, as these were incorporated in the computational domain of ECLIPSE. Additional capabilities also still need to
be embedded into IC-FERST, such as the incorporation of the fluid density and viscosity variations with pressure and the consideration of the volume factors, in order to enhance its competitiveness with existing commercial reservoirs simulators such as ECLIPSE.