Vertical two-phase flow studies and modelling of flow in geothermal wells

Abstract

A detailed study of two-phase pipe flow with particular reference to vertical steam-water flow is presented. An analysis of the application of current techniques in two-phase flow to geothermal flow is described, and an indepth study on the parameters influencing geothermal wellbore flow is included. Five available geothermal wellbore simulators which use different methods were critically reviewed, tested and compared. No single simulator was found suitable to deal with all of the well data used. In this study mechanisms of two-phase flow are modeled; these are based on sound analytical and experimental observations wherever possible, preferably with large pipe diameter, steam-water applications. A new flow regime map was developed and is presented. It was decided, with the support of experimental data, to evaluate pressure drop in churn flow in the same way as that in slug flow. It is also suggested that heat transfer analysis in a geothermal well is simplified if the thermal resistance of the fluid is ignored. An extensive programme was carried out to study the dynamics of flow in a well with particular reference to slotted liners. Flowing well data from different production wells were studied to obtain representative roughness values for the slotted liner when it is assumed that the total flow is inside the liner. Laboratory tests using air as medium and a computer program developed to study flow in the slotted liner part of the well showed that the fluid is actually distributed in the slotted liner and in the annulus between the liner and the open hole. The amount of flow in each duct depends on the geometry of the flow, the roughnesses of the slotted liner and open hole, and also the fluid properties. The pressure gradient in the annulus and inside the liner was, however, found to be the same except near the inlet and exit zones. Field injection tests gave complementary results for temperature, pressure and flowrate runs. It was possible to identify loss zones using the recorded data. It is shown that the slotted liner can be assumed to have a roughness equivalent to a smooth pipe when the total flow is assumed to be inside the liner. The injection tests also demonstrated how the roughness of the production casing is affected by deposition. It is demonstrated that a linear relationship between the mass flow and the pressure draw down in the reservoir applies at low flows only. At high flowrates turbulence and other effects become important and the draw down relationship is no longer linear. New computer codes were developed. A simulator for liquid or two-phase flow is introduced in two versions to allow flexibility. The first version (WFSA) calculates up or down a well, includes dissolved solids and heat loss and can allow multi-feed well simulation provided information on the feed zones is available. The second version (WFSB) can handle gases, dissolved solids and heat loss. The simulator was validated with a variety of well data including wells with gases and dissolved solids, and produced good results. Better results in the simulation of wells with gas were obtained when the mass fraction of gas in the vapour phase, was defined as a ratio of pressures instead of ratio of densities (Dalton's Law). The presence of gas even in small amounts (< 0.5 % by weight) was found to affect simulation. Also a wellbore simulator (STFLOW) was developed to model wells (including multi-feed wells) which produce dry steam. Comparison with other methods and with field data showed good results. Exercises using this simulator for a hypothetical two-feed well proved very helpful in understanding the sensitivity of the output to changes in parameters, such a permeability and depth, of the feed zones.