Contact: Wendy Soll, Team Leader MS F665, Los Alamos National Laboratory Los Alamos, NM 87545 Voice (505)665-6930 FAX (505)665-3687 wes@vega.lanl.gov

Pore Scale Processes

One of the continuing efforts in EES-5 has been the modeling and analysis of porous media systems involving multiple fluids. Such systems are of interest to both the groundwater community, with application to remediation of immiscible contaminant, and the petroleum industry, for application to enhanced oil recovery and remediation of oil fields. The complexity of porous media systems and the difficulty in accurately modeling them with the traditional averaging (Darcian) approaches has led us to look for alternative means of predicting fluid movement. In this effort we take a pore scale approach, to directly account for the fundamental physical processes that affect fluid behavior. This is a "bottom up" approach in which the porous media are discretized to sub-pore resolution, fluid movement and interaction with the solid is modeled within the finely discretized pore space, and continuum scale parameters, such as velocities, pressures, permeabilities, etc., are extracted through appropriate averaging techniques.

The pore scale model is based on a numerical technique called lattice Boltzmann (LB) or lattice gas, that is able to incorporate the fundamental physics and exactly solve the Navier-Stokes equations. The LB approach allows us to incorporate basic fluid and rock properties, such as pore space geometry, wettability, interfacial tensions, with fundamental fluid dynamics, and obtainsolutions for the flow equations in a very complex pore network. This approach offers three major benefits over traditional techniques: we can relate porous media and fluid properties to the measured values, we can improve predictions of fluid behavior over those from traditional macroscale approaches, and we can obtain these benefits with less cost and effort.

---

Figure 1a

Figure 1b

Figure 1a and b Oil displacement by water flood. In a two dimensional pore space. Water is shown in red, oil is shown in blue, and the porous medium is shown in black. a) Nca = 10-2 and b) Nca = 10-4

The LB modeling approach is a "discrete fluid" approach that models the fluids as moving "fluid packets" on a regular lattice discretizing the system. These fluid packets move and interact on the lattice, with a set of rules that governs what happens when the packets collide at the nodes of the lattice. With the proper choice of geometric lattice and interaction rules these systems can reproduce flow and transport of highly complex systems, with multiple fluids and heterogeneous porous media. The algorithms used by the LB model are readily parallelizable, and are extremely efficiently implemented on the massively parallel Connection Machine-5 (CM5) computer here at Los Alamos.

Our model is capable of simulating fluid flow in three-dimensional porous media of arbitrary pore geometry and rock surface characteristics, with up to three fluids having arbitrary density, viscosity, and interfacial tension ratios. Thus, the model has the flexibility to reproduce a majority of the systems of interest in groundwater or petroleum systems. Using the CM5 we regularly use porous medium domains of dimension 128x128x128 and 256x256x256. The large geometries are taken from actual sandstones and correspond to a cubic sample of dimension 1.28 mm on a side (geometries provided by Mobil Oil). We have also utilized high resolution two-dimensional systems (1024x512 or 2048x1024) to compare against visualization experiments being done in conjunction with this work. The figure on the cover shows water (blue) displacing oil (yellow) from a producing reservoir sandstone. The system dimensions 113x145x145.

Figure 2. Relative permability-saturation-relationship for simulation in Figure 1a).

We have been running simulations to compute relative permeability - saturation relationships, and to study the effect of rock surface wettability and flow rates on relative permeabilities and displacement efficiencies. These simulations and calculations are currently being applied to improving predictions of oil recovery efficiency. Figures 1 and 2 demonstrate some of the simulations we are doing using the LB pore scale model. Figure 1 demonstrates the effect of capillary number (the ratio of viscous forces to capillary forces) on displacement efficiency. In this case the pore space is a high resolution digitization of a two-dimensional manufactured pore space. Note that the higher capillary number flow (1a) results in much better displacement than the lower, capillary dominated, flow (1b).

---

We are currently expanding our pore scale studies to other areas of concern in porous media systems. We are studying the interaction of fracture and porous matrix on flow in fractured porous media. The fundamental approach of the LB code is such that in implementing a fractured pore space the flow processes are inherently captured, and only a change in the pore geometry is required. In a collaborative effort the LB model has been extended to account for reactive processes, such as dissolution, precipitation, and sorption/desorption.

EES-5 | EES Division | LANL | DOE
Phone Book | Search

L O S   A L A M O S   N A T I O N A L   L A B O R A T O R Y
Operated by the University of California for the US Department of Energy