Fine-Scale Simulation of Sandstone Acidizing

Abstract

Abstract The current state of the art of sandstone acidizing modeling considers heterogeneities only in a very gross manner. The most sophisticated design models treat the formation as a series of layers with constant properties (mineralogy, permeability, etc.) in each layer. A radial variation in formation properties may be considered as a method of simulating the damaged region, but there is typically only two discrete regions considered – the damaged zone extending to some assumed radial distance, and the unaltered formation beyond this distance. However, sandstones invariably have small – scale heterogeneities in mineralogy and flow properties that may cause the effects of injected acids to differ greatly from what is predicted by a model based on a homogeneous formation. Matrix acidizing of sandstone is influenced by very small-scale variations in the permeability field. Preferential flow of the acid through higher permeability pathways in the matrix allows much deeper acid penetration and better overall permeability response than is predicted by any current model of sandstone acidizing. This may be particularly important for high concentration HF treatments that are currently being applied, where channeling of the acid may create wormhole-like structures. In addition to variations in the permeability field, variations in the mineral distribution affect the stimulation achieved with acid. For example, a sandstone containing clay-rich streaks may experience a large permeability increase as the clay is dissolved by acid. We have developed a fine-scale model of the sandstone core acid flooding process. The model divides the domain into grid blocks on the order of 0.1 inches on a side to simulate the fine-scale structure of sandstone. A highly variable matrix domain was generated by using standard geostatistical techniques, and the acid and mineral balance equations allowed us to predict the changing permeability field as acidizing proceeds. Permeability response to acidizing is predicted in which not only the porosity, but also the mineralogy, tortuosity, and statistical parameters of particle size are considered. The initial porosity and mineralogy field can be generated in a correlated manner in three dimensions; thus, a striated or laminated sandstone can be simulated by the model. Application of the new model to typical acidizing conditions shows that acid tends to channel through a heterogeneous sandstone, with the most efficient acidizing occurring when the rock has a layered structure. Layering is simulated by assuming a correlated permeability field in the main flow direction, as occurs in sandstones having horizontal laminations. The model shows that acid can stimulate the matrix permeability two to three times farther into the rock than would be predicted with a standard acidizing model. The effects illustrated for acid penetration would also apply to other injection processes that rely on transport and chemical reaction in the matrix.