Large scale production of heavy oil and bitumen is possible using the steam-assisted gravity drainage process. This process involves the use of horizontal production wells combined with either horizontal or vertical injectors. One application of the process is to enhance the recovery from depleted cyclic steam stimulation projects. In this application, horizontal wells are drilled at the base of the reservoir and the existing vertical wells are used for injection.
In this paper, the growth of steam chambers around a vertical well, as heated oil drains to various production well types, have been studied in scaled physical model experiments, using a three-dimensional cylindrical apparatus.
Canada posesses vast deposits of bitumen and heavy oil. With the development of horizontal drilling techniques, the steam assisted gravitv drainage (SAGD) process(1,2) has become a promising method of recovering these resources. In this process continuous injection of steam creates a "steam chamber", Once the steam chamber has developed, oil flows from its perimeter to a production well located below. The heated oil is driven by the force of gravity. Reservoir contact and production rates of the otherwise slow process are improved by using horizontal production wells. Growth of the steam chamber continues as time progresses (3).
At present, in Canada, bitumen is recovered by the cyclic steam stimulation process [4,5,6]. Relatively poor recovery figures for these methods make it desirable to develop methods for improved recovery. One idea is to employ the SAGO principle by drilling horizontal production wells near to the base of the reservoir and using existing vertical wells for injection. It is, of course, also possible to use SAGO from the start as has been done at the AOSTRA UTF project (7).
In this paper, experimental observations of steam chamber growth are discussed and production performance of a horizontal well is developed by correlating experimental data from planar, horizontal and vertical production wells. A subsequent paper is planned which will involve a theoretical analysis and correlation.
A new, cylindrical, three-dimensional reservoir model (shown in Figure 1a) was used in the experimental study. Criteria developed ea r1ier  ca n be used to scale the model to field conditions. For example, operating the model for one hour is approximately equivalent to 1.4 years of operation in the Cold Lake field. The model was built from a 20 inch 00 schedule 30 XS steel pipe, 9 inches in height.
To avoid downward heat conduction once the steam chamber reaches the vessel wall, the inside of the cylinder was insulated, using a concrete made with a granulite aggregate.
The top and bottom of the vessel were made from 1–1/4 inch thick phenolic. One quarter inch fittings in the top lid were used to secure the steam injector at the center, as well as 5 mm glass tubes which house four and sometimes five T-type thermocouples along perpendicular diameters, as shown in Figure 1.
The model was filled with 2 mm diameter glass beads.