A Generic Approach to Modelling Fractures and Fluid Flow in Naturally Fractured Carbonate Reservoirs and Its Impact on Development Strategy

Jean-Marie Questiaux, Nicolas Ruby, and Gary Couples
Institute of Petroleum Engineering, Heriot-Watt University, Edinburgh, Scotland, UK

A high percentage of the world’s oil reserves are found in fractured reservoirs, which are characterised by production problems such as high initial flow rates, rapid decline, early gas or water breakthrough and ultimately low recovery. However, properly developed fractured reservoirs can have comparable recoveries to conventional reservoirs (Jack Allan et al, 2003).

This study focused on modelling fractures and their fluid flow behaviour in low porosity and low permeability reservoirs, typical of some of the largest carbonate fields in the Middle East. In these fields, the production is controlled primarily by conductive fractures providing the main fluid-flow pathways (Type II fractured reservoirs) and seldom achieves recoveries above 20%.

Generic models were constructed on the basis of an existing giant Middle Eastern field located in the foothills of the Zagros Mountain Fold Belt. This allowed a comparison between the model results and the field’s actual dynamic information from DST’s, Production Logs (PLT), Production Rates, Productivity Indices (PI), Mud Losses and water or gas breakthrough. The reservoir is in the Oligo-Miocene Asmari Limestones, between 400m and 450m thick, and with low matrix poroperm characteristics. The trap is an elongated tight asymmetric anticline, with a large oil leg topped by an important but aerially limited gas cap over the centre of the field. Recovery to date from this field is less than 11% after 40 years of production by essentially natural depletion.

Discontinuities such as fractures are a fundamental consequence of rock deformation controlled by the tectonic setting. Flexural slip folding is the believed mode of deformation in the foothills of the Zagros Mountains, and has been adopted to model the fracture system in this field. Our understanding of the geomechanics of flexural slip folding and the spatial and temporal distribution of the resultant discontinuities forms the base for modelling fractures in this study.

In flexural slip model, domains of fracture inducing strains are partitioned into stratabound mechanical units of competent rocks, separated from each other by surfaces where less competent rocks allow slippage to occur between the mechanical units (Couples et al, 1998). The distribution of these strain domains is controlled by the mechanical unit’s position in the fold (hinges or limbs), the amount of fold curvature and the degree of differential slippage along the bounding slip surfaces. The fracture orientation, intensity and characteristics in each mechanical unit are likewise related to the strain domains, but also depend on the thickness of the Mechanical Units, the lithology type and rheology, and the present stress state.

The reservoir model is constructed as a series of stacked mechanical units with low poroperm reservoir matrix, each unit cut by its own stratabound conjugate set of fault corridors 1 to 10’s of metres wide and 100’s of metres to kilometre long forming lozenges 100’s m² to Km² in surface area. The effective permeability of the matrix outside of the fracture corridors will depend on the development (at varying scales) of microfractures, hairline cracks (cm) and macrofractures (m). Dolomitisation will naturally also affect matrix permeability, but this was not considered in this study.

The fracture corridors act as the main drains and conduits for the oil stored within the lozenges. The productivity of the corridors depends on their permeability and storage capacity, while sustainability of production will on the other hand depend on the extent of the fracture corridors and the volume of matrix reservoir it drains.

Not withstanding the comments above, the distribution of fracture corridors and the poroperm characteristics both inside the fracture corridors and matrix were stochastically modelled in each Mechanical Unit, using a commercial geomodeller. The result is a model of 10 Mechanical Units (layers) 30m to 45m thick, each with its own random distribution of fault corridors cutting across a low poroperm matrix.

The fluid flow in the constructed fracture models has been evaluated in a single porosity phenomenological sector model, using a commercial 3D dynamic modeller and the results compared to the reservoir and well behaviour of the actual field. The results from the reservoir simulations approximate well the reservoir behaviour, including the low recoveries. This does not however confirm the model’s validity, which will require further refinements and integration of the large well dataset from this and analogue fields before reaching that stage.

Using the dynamic model does however allow the investigation of the optimal completion and production scenarios likely to improve productivity; maximise recovery and minimise risk of early gas and water breakthrough. This entailed assessing the performance of various production well trajectories, comparing the efficiency of completing wells in the matrix rather than in the fracture corridors, determining the optimum positioning of wells relative to the OWC, evaluating the impact of operating parameters and finally investigating secondary recovery methods (water or gas injection).

Results from these sensitivities clearly show that recovery can be significantly improved by up to a factor of three to around 30% by water injection support and optimal well placement. This has very significant economic implications, which is confirmed from simple NPV calculations for optimal development scenarios against the present base case of natural depletion.

Copyright © 2004. The American Association of Petroleum Geologists. All Rights Reserved.