Abstract
This paper describes THERMS, a new three-phase, black-oil simulator for in-situ combustion processes. The model also can simulate steam and hot-water injection operations. To simulate field-scale in-situ combustion projects, THERMS uses experimental data obtained from laboratory combustion-tube runs. Detailed kinetics and thermodynamic relations are not used. The model treats the burning zone as a moving front. The front is viewed not only as a moving heat source, but also as a displacement pump enhancing oil flow. This approach results in a new algorithm. The model has been designed with strongly field-oriented features. Its utility is illustrated by a history-matching study of the South Belridge, CA, dry-combustion pilot test.
Introduction
A key factor in the numerical simulation of in-situ combustion processes is the representation of the burning-zone effects. These effects can be represented in two different ways. One way is to describe the complex physiochemical phenomena observed in the zone with a set of elementary kinetic expressions. The other way is to approximate the overall effects of the observed phenomena with a few measurable parameters. Most of the published combustion simulators follow the first representation. They use finite-difference methods to solve the set of conservation equations numerically. This results in the division of the reservoir into cells or blocks. Each block is treated as a continuously stirred tank reactor (CSTR). Therefore, the reservoir is converted into a network of CSTR'S. Within a CSTR, combustion kinetics and thermodynamic phase-equilibrium relationships are treated. These relationships include many parameters (e.g., activation energy, frequency factors, and reaction order) that are either unobservable or very difficult to determine through routine laboratory experiments. A major shortcoming of this approach is the high level of uncertainties associated with the values of the parameters. Another significant shortcoming results because the thickness of the burning zone is considerably smaller than the size of grid blocks normally used. Because of this, existing simulation methods do not yield accurate distribution of solution parameters, such as temperature. The combustion kinetics use the Arrhenius-type expressions, which are exponential functions of temperature. Thus, inaccurate temperature distribution leads to significant errors. The second approach has not been explored previously. The key idea is to specify an amount of oil to be burned per unit bulk volume of formation. The burning zone is considered a moving interface. The complicated relationships of the combustion kinetics and phase equilibria are replaced by combustion parameters that are relatively easy to determine. Prats et al. and Dietz and Weijdema used a simplified version of this approach to estimate behavior of the burning front in a simple reservoir system without multiphase flow. While the second approach simplifies the kinetics' calculations, it requires an algorithm to link the burning zone to the unburned portion of the reservoir. This algorithm permits the use of the finite-difference methods of three-phase flow. In developing the algorithm, two simplifying assumptions are used:oil is one component and is described by the normal PVT relationships, andinstantaneous and complete combustion occurs whenever oxygen and oil meet.
The second assumption permits the problem to be posed as a moving-boundary problem. The enhanced oil recovery resulting from the burning zone is approximately that resulting from the burning front. The burning front is viewed as a moving heat source and as a displacement pump enhancing oil flow.
SPEJ
P. 271^
Knowledge Discovery for Classification of Three-Phase Vertical Flow Patterns of Heavy Oil from Pressure Drop and Flow Rate Data