During the past decade a number of pipelines have been laid through which oil and gas are flowed simultaneously. In certain areas such as swamps and offshore locations, this method of transportation has been favored. Whether the advantages gained from separate lines offset the additional cost of those as compared to a single line carrying both is still a moot question.
Another problem which is encountered in two-phase flow is the feasibility of injecting a liquid drying agent, such as diethylene glycol, into natural gas lines in an effort to reduce hydrate formation. In pipeline design the prediction of the increase in energy loss and liquid holdup are questions which plague the engineer and will remain unanswered until the mechanism of two-phase flow is better understood.
It is essential that lines carrying two phases be of sufficient diameter so that the pressure drops will not be too high and yet their size will be small enough so that the cost will not be excessive. It is well known that the energy losses experienced in two-phase flow are ordinarily higher than those for single-phase flow. This is due not only to the irreversible work done by the gas on the liquid as it is dragged along but also to the reduced flow areas resulting from the presence of two fluids. No completely satisfactory means has yet been found to calculate the pressure drop resulting under a given set of conditions. Also, the dependence of the various flow types upon flow rates and other physical properties is not well understood. As two-phase flow now is being used to a greater extent in industry, the need for additional experimental information has increased. It was with this thought in mind that the present investigation was undertaken.
Some of the early experimental work was reported in 1939 by Boelter and Kepner. They studied two-phase flow occurring in a fuel distributing system for heaters in California orchards, and their data cover flow in 1/2 and 3/4 inch metal pipe of 50 ft lengths. Although they presented no correlation, it was suggested that the pressure gradients were functions of various dimensionless groups.
In 1944 the first general correlation was developed by Martinelli, Boelter, Taylor, Thomson and Morrin. A further analysis was made by Martinelli and co-workers covering the viscous region in 1946. Their tests were made in tubes of two to 50 feet in length ranging in size from capillary tubing to one inch pipe.