Calibration of fluid models using multiphase well test data for improved history matching

History matching of reservoir performance is difficult due to the uncertainties associated with reservoir properties. These uncertainties include the variety of geological model and fluid model. Usually, reservoir engineer changes geological properties to match history data, but ignore the changes of fluid properties, which may have a big influence on history matching.Considering a mature field entering high water-cut stage, for many infill wells, the well productivity can be obtained by stabilized well testing, but there is a phenomenon for drawdown (DD) test where the bottom-hole pressure (BHP) may increase with time when the water-cut reaches some value. In this situation, many reservoir engineers think this phenomenon was due to the change of production rate, the change of skin factor near wellbore or the interference of multi-wells. This paper excludes the interferences of these reasons, the objective of research focus on multiphase flow and explains the phenomenon from the viewpoint of multiphase flow.It is well known that the change of pressure on late time of well testing is closed to constant in closed system or closed to zero in constant pressure system. But under transient state, steady-state or semi-steady-state conditions, if the bottom-hole pressure goes up, then the DD data may not be used for well test analysis or reservoir engineering analysis. This paper presents study results on wells producing a multiphase reservoir by numerical well test model. According to Darcy's Law, a new theoretical expression was derived in this situation, in which the derivative of bottom-hole flowing pressure is a function of integral of total mobility, which can be approximatively replaced of the average total effective mobility. Using this equation, the reason why well bottom-hole pressure increasing with time may be reasonably explained.According to the results of this study, under multiphase flow condition, the BHP increasing is due to the changing of the fluid total mobility, in this paper, it was further demonstrated that the data of this type can be used to adjust fluid model and improve history matching. © 2010.