Accounting for Silica Precipitation in the Design of Sandstone Acidizing

Abstract

Summary Mud-acid treatments normally are designed by approximating the complex mineralogy of a sandstone with "lumping" procedure. Minerals are classified as either fast- or slow- reacting and the rates of their reaction with HF are determined by analysis of acid effluent from acidized core plugs. For most treatments carried out at modest temperatures and reasonable rates, this approach is satisfactory. In this paper, we show that at higher reaction temperatures, the simple two-mineral dissolution model does not apply because an intermediate product of the HF reaction with quartz, feldspars, and clays (H2SiF6) begins to react further with both clays and feldspars. This new reaction must be included to model data. The additional reaction, not observed at lower temperatures, has important consequences. For example, the acid injection rate is no longer critical. The analysis presented here shows that retarded acids are unnecessary. Contrary to previous concepts, mud acid itself provides a deep-penetrating capability. This surprising result may account for the high percentage of successful treatments even when this treatment is carried out under a wide range of conditions. Introduction Sandstone acidizing is a complex but widely applied and often successful well-stimulation process. A sandstone is an intricate composite of many minerals exhibiting a wide variety of morphologies. The HF reaction rate differs widely from mineral to mineral because of variations in the intrinsic reaction rates and the area of contact with the pore fluids. If the main reactions are rapid compared with the fluid flow rate (large Damkohler numbers), then a local equilibrium will apply and the process will be characterized by thermodynamic variables.1 In this case, the minerals initially present dissolve in a well-defined sequence while new minerals may precipitate. By this dissolution and precipitation process, zones of different mineral compositions develop and move through the porous medium at a rate strictly determined by the total number of PV's injected but independent of injection rate. At local equilibrium, Walsh et al.1 reported that an essential feature of even a well-designed sandstone treatment is the dissolution of aluminosilicates with the attendant precipitation of some form of monosilicic acid [Si(OH)4]. There are various possible forms of the precipitate, but here we call them silica gel. At the other extreme are the conditions for which at least some of the reactions are slow compared with the flow rate (small Dakohler numbers). Many models developed to represent the complexities of sandstone acidizing ignore those reactions, which lead to precipitation, and essentially consider only dissolution rates. The most widely used model divides the minerals that are found in sandstones into two categories-fast reacting (feldspars, authigenic clays, amorphous silica, etc.)2–5 This model, which is discussed in more detail here, has become known as the two-parameter model.2–4,6,7 There is little evidence defining the conditions where the rates of certain crucial reactions taken to be negligible in the two-parameter model begin to become significant. Recently, however, Bryant7 interpreted the failure of the two-parameter model to fit all three high-temperature core-plug acidizing experiments reported by Lindsey8 as an indication that the rates of certain reactions that appear at local equilibrium but are neglected in the two-parameter model now must be considered. The purpose of this paper is to examine Bryant's hypothesis and to define its practical implications. Reaction Kinetics The reactions for the slow- and fast-reacting minerals areEquation 1 and 2 The vk are stoichiometric coefficients. These reactions have been found to approximate first-order kinetics2,4,5 so that the following kinetic model is used:Equation 3 Where Cmin,k=concentration of accessible mineral k (an inaccessible mineral is not contacted by the pore fluids during the acidizing process2). We will show that this model fits the available data at low reaction temperatures where the reaction of the fluosilicic acid, H2SiF6, produced by Eqs. 1 and 2 with Mineral 1 is slow. At higher temperatures, reactions rates increase and the system tends to approach local equilibrium. The studies reported by Walsh et al.,1 which assumed local equilibrium, represent the limit of fast reactions. Walsh et al., found that large amounts of Si(OH)4 precipitate and, once formed, are difficult to remove by continued HF injection. In fact, for total acid volumes corresponding to those normally associated with acid treatments, this precipitate probably will remain in the near-wellbore region, and depending on where it is deposited in the pore structure (i.e., pore throats vs. pore bodies), it may be damaging. Thus, the issue is defined clearly. For higher temperatures (deep formations), H2SiF6 formed by the reaction of HF with Minerals 1 through 3 may itself react to a significant extent according toEquation 4 This reaction is a simplification of the comprehensive set of reactions that Walsh et al. Considered but is included in Bryant's model. Labrid,9 Shaughnessy and Kunze,10 and Crowe11 experimentally confirmed the existence of the reaction. The mechanism of fluosilicic acid reaction with clays or feldspars has not been studied, but considering it to be first-order seems reasonable. Thus, the rate of appearance of fluosilicic acid per unit volume is given byEquation 5 Note the Eqs. 1 and 2 produce fluosilic acid and must be included to obtain the net production rate.