Emulsion Deposition in Porous Media: Impact on Well Injectivity

Emulsions flow is encountered in various pratical applications such as enhanced oil recovery (EOR) processes. The occurence of emulsions can have a detrimental impact on well productivity/injectivity and hence, on the enconomic of many EOR operations. However there is still a lack of clear understanding of how they really behave inside a porous medium. The present work aims at giving new insights into emulsion deposition mechanisms. Different experimental conditions are explored to study the deposition of "stable" oil-in-water (O/W) emulsions inside a porous material. Porous media made of sharp-edged Silicon Carbide (SiC) grains of 80 m in diameter and dilute and stable dodecane emulsions in brines were used. The ratio of pore to particle size (jamming ratio) is set high enough to ensure in-depth propagation and hence, an in-depth deposition on pore surface without any other retention mechanism such as straining and plugging. In this study, the effects of salinity and flow rate on droplet deposition are highlighted. At low salinities, the deposition and the induced permeability reduction are negligible. As SiC is negatively charged, this indicates that oil droplets bear a negative surface charge, even if they are stabilized by a non-ionic surfactant. Increasing the salinity induces a better deposition efficiency and a concomittant increase in permeability reduction by screening surface charges. Moreover, the piston-like displacement of the emulsion front entails that the available surface of each section is almost entirely saturated before the next one is reached. Thereby, the deposition is almost uniform along the core. The breakthrough curves exhibit a delay that increases while increasing the salinity and a plateau value near saturation (C/C0=1). This reveals two different deposition steps characterized by two depositon kinetics which will be described according to well known scaling power law for colloidal spheres. At high salinity and at the lowest flow rates (in the convection-regime), the maximum permeability reduction reached is in excellent agreement with the predicted value from Random Sequential Adsorption (RSA) model. That was initially developed for hard spheres. That confirms that, for "stable" O/W emulsions, droplets deposit as a single layer probably because of steric hindrance. Interestingly, the permeability reduction reaches its maximum at breakthrough where the surface coverage is close to 54%, as predicted from the RSA theory. However, although the permeability impairment is in agreement with the RSA theory, the total deposited amount of oil is higher than expected from the RSA model predictions. After the breakthrough, the droplet deposition continues increasing but at a very small rate during the aforementioned second deposition step, and this, with no significant impact on permeability. It infers that the deposit may get denser on grain surface owing to droplets distortion. The final surface coverage is thus higher than the upper limit depicted in the RSA theory. Finally, from the desorption and the mean hydrodynamic thickness of the deposition layer, we get some experimental clues that, the droplets are likely to deposit individually and not to coalesce to form a film. Introduction Emulsion flow and retention have attracted a growing interest since 30 years as they are involved in many oil processes. Several models (Mc Auliffe, C. D., 1973; Devereux, O. F., 1974; Alvarado, 2 S.Buret, L.Nabzar and A.Jada IEA Collaborative Project on EOR 30Annual Workshop and Symposium 21-23 September 2009, Canberra, Australia D. A. and al, 1979) have been developed over the years but none of them describes the whole damaging process caused by emulsions. The most accurate and reliable model is based on the deep-bed filtration (Soo, H. and al, 1986) and accounts for the residual permeability reduction and the shape of the breakthrough curve. Experimental study have mainely been carried out on macroemulsions which could be useful as plugging or EOR agents (Romero, L. and al, 1996; Khambharatana, F. and al, 1998). In most of the cases , the authors explain the permeability reduction by droplets straining in pore throats smaller than their size. This same conclusion is sometimes found even if miniemulsions are under consideration (Soma, J. and al, 1995) because of droplet destabilization inside the porous medium. In fact, permeability impairment involves four successive steps (Roque, C. and al, 1995) : surface deposition, pore bridging, internal cake formation and finally external cake formation This paper focuses on the first stage of the impairment process. To that point, small sterically-stabilized droplets of oil are injected in highly permeable cores. Some theoretical results, useful for the interpretation of results, are exposed in the first section. Then, the material and procedures are described and experimental results are given and discussed in the last part. Theoretical background Modeling of surface deposition In our experimental conditions, the Jamming ratio (Jr), ie the mean pore size to the mean droplet size ratio, is above 40 what avoid any pore bridging and any possible external cake formation. Theoretically, Jr has to be higher than 3 (Barkman, J. H. and al, 1972) but the size ditributions of both pores and droplets increase this value. However, the deposition will here only result from adsorption on grains surface. Colloidal approach To describe particle surface deposition, the grains of the porous media are considered as spherical collectors of radius ag and the incoming oil droplets are assumed to be carried in the vicinity of the collector by the fluid at the interstitial velocity u (Fig. 1). The particle deposition on the surface depends on the coupling of the transport mechanism to the surface and the physicochemical interactions. Figure 1: Representation of a spherical collector grain and an incoming particle, with the stream lines and the shape of the diffusion layer. In this particular approach, the relevant dimensionless number describing the particle deposition is the Grain Peclet Number, Peg, which describes the balance between convective and diffusive fluxes: Emulsion Deposition in Porous Media : Impact on Well Injectivity 3 IEA Collaborative Project on EOR 30Annual Workshop and Symposium 21-23 September 2009, Canberra, Australia D ua Pe g g = where D is the particle diffusion coefficient, p a 6 kT D πμ = . Depending on the value of Peg, different deposition regimes occur (Chauveteau, G. and al., 1998; on emulsions see Rousseau, D. and al., 2007). Convection-diffusion regime is the only regime further discussed according to Table 1. TABLE 1 GRAIN PECLET NUMBERS Range of Grain Peclet Numbers explored Upper limit of convectiondiffusion regime 1338 to 5.35x10 1.19x10 Convection-diffusion regime Solving the convection-diffusion equation gives different relationships for the collection efficiency, η, versus the Grain Peclet Number Peg depending on the existence of an energy barrier. is defined as the ratio between deposited and incoming particle fluxes. In the convection-diffusion regime, the relationship η vs. Peg is expressed through the general scaling law form α − ∝ η g Pe If double layer interactions can be neglected and if Van der Waals interactions exactly counterbalance the hydrodynamic friction then = 2/3 (Levich, V. G., 1962). This regime is named DLD for Diffusion Limited deposition. If, at the opposite, the energetical barrier is significant between the particle and the collector then =1 (Spielman, A. L. and al, 1974). This regime is named RLD for Reaction Limited Deposition. Thus, all the experimental values of lie between 2/3 and 1. Boundary of the convection diffusion regime It is generally admitted that the above relationships are relevant for Grain Peclet Number between 100 and (ag/ap) (Prieve, D. C. and al, 1974). Some results of the RSA (Random Sequential Adsorption) theory This model is largely used to describe the deposition of suspensions made of hard spheres. Experimental study and simulations (Lopez, P. and al, 2004) show that, under these hypotheses, the surface coverage, , can't exceed 54% of the total avalaible surface and that the permeability reduction Rk is given by: