This paper describes comparisons between the prediction of high resolution simulation and the results prediction of high resolution simulation and the results of detailed laboratory measurements of miscible flow in a heterogeneous rock slab. The permeability distribution was characterised using minipermeametry and an X-ray radiography technique was developed to obtain quantitative concentration maps. Displacement experiments were conducted at four mobility ratios. Both stable and unstable flows were investigated, enabling the effects of physical dispersion and viscous fingering to be quantified. Excellent agreement was obtained between simulation and experiment without the use of history matching. The simulations predicted the development of individual viscous fingers as well as the correct effluent and recovery profiles. This work has validated characterization techniques and demonstrated our ability to simulate single phase flow in a heterogeneous rock. The experimental results also provide a set of data for use in bench-marking other provide a set of data for use in bench-marking other simulation packages.
The efficiency of a field-scale miscible gas project will be affected by viscous fingering. The interface between the less viscous gas and the more viscous oil is unstable. Small perturbations along the interface result in the foration of viscous fingers. These channel through the oil resulting in early solvent breakthrough at the production well and a reduced sweep efficiency. It is clearly important to be able to model the effects of these instabilities in order to predict breakthrough time and the amount of oil recovered.
Conventional reservoir simulation cannot resolve the details of viscous fingering, so empirical models have been developed to represent their average effects. The methods described by Koval (1963) and Todd and Longstaff (1972) are incorporated in a number of simulation packages. The empirical model parameters need to be calibrated, ideally by comparison with a model of the detailed flow on a sub-grid block scale.
High resolution simulation can be used both to study the detailed flow and to calibrate these empirical models. It is important to ensure that the numerical methods and the physics used in high resolution simulation are correct. This can only be achieved by comparing the predictions of simulation with results from a series of well-characterized experiments.
Recent work has demonstrated our ability to accurately simulate unstable miscible displacements in a homogeneous beadpack. Christie and Bond (1987) obtained good agreement between high resolution simulation and Blackwell's (1962) experiments. Christie and Jones (1988) found similar excellent agreement between simulated and observed production rates and average concentrations for a horizontal bead-pack. Most recently Christie, Jones and Muggeridge (1989) have shown that the same 2D high resolution simulation program can accurately predict the transition from program can accurately predict the transition from viscous dominated flow to gravity dominated flow as long as the experimental flow is also 2D.