Single Porosity Simulation Of Fractures With Low To Medium Fracture To Matrix Permeability Contrast

Abstract

Abstract The use of single porosity models to simulate fracture flow is a challenge for simulation engineers. The main challenge remains in the speed at which water advances in the fracture system with little production from the producers that intersect the fractures. In single porosity modeling, water has to displace the oil ahead of the flood front to reach a producer. Due to the high matrix to fracture storage ratio, this usually results in over-prediction of the produced oil volume prior to water breakthrough. One typical solution to this problem is the use of local grid refinement, where an engineer will represent the fracture or fracture corridor with a "thinner" line of cells. Another alternative solution is to modify the relative permeability curves of oil and water, to slow the oil movement and speed up the water movement in these cells. Paul van Lingen et al (2001) presented a technique that relies on modifying the relative permeability curves to produce "pseudo relative permeability" curves for the grid blocks containing fractures, without the need for grid modification. This method reflects a considerable advancement in the modeling of fracture system behavior in single porosity systems. It allowed to model faster breakthrough times and higher water production rates after breakthrough. This paper describes a modified pseudo relative permeability correlation, which is based on the van Lingen et al method, that improves the water breakthrough time and water cut predictions, especially in the low to medium fracture-to-matrix permeability contrast cases. The new method restricts the pseudo curves to the widely known straight line curves that are typically used to represent fracture flow in dual porosity modeling. The modified correlation was numerically tested with various permeability contrast cases and was compared to a dual porosity model. The method is currently being used for modeling a large Middle East oil reservoir. Introduction In characterizing reservoirs, Narr et al. (2006) stated that "all reservoirs should be considered fractured until proven otherwise." Whether to designate a reservoir as a naturally fractured reservoir or not will depend on the degree to which fractures affect reservoir performance. In cases where fluid flow from matrix to fracture occurs at a rate that is significantly slower than flow within the fractures, it is necessary to use a dual media approach to the flow simulation (Narr et al. 2006; Daily and Mueller 2004). In other words, a fractured reservoir, from the fluid-flow point of view, means that the total volumetric flux in fractures must be substantially larger than the total volumetric flux in the matrix blocks (Salimi and Bruining 2008; Nelson 2001). A significant advancement in the area of modeling fractured reservoirs began after Warren and Root's (1963) original work, which developed an idealized model representing fractured reservoirs (Fig. 1), and applied it to well test analysis. Their work was based on the initial work by Barenblatt et al in 1960 who introduced the dual porosity concept for modeling fractured reservoirs (Salimi and Bruining 2008). When dealing with dual porosity modeling, one has to deal with more than twice the data that is needed for a single-porosity model. Further complicating the situation is the fact that there is high uncertainty associated with the fracture system description and characterization, particularly with respect to fracture length and connectivity, and the impossibility of measuring parameters which directly dominate reservoir behavior (Reiss 1980). Furthermore, the computational demands of dual porosity models are much higher than single porosity models. When dealing with giant oil reservoirs, commonly found in the Middle East, the size of the reservoir models become enormous, and the computational demands become prohibitive.