| The effect of well trajectory on two-phase flow behavior in horizontal gas and condensate wells was experimentally studied. Experiments were conducted in a facility with a 2-in. ID and 62.6-ft. long lateral section and a 41-ft. vertical section, which mimics a horizontal well. Four well configurations were considered; namely, toe-down, toe-up, one-undulation with a sump and one-undulation with a hump. The experimental program was divided in two stages. The first stage was the static testing which simulates the well startup and evaluates the effect of the well trajectory on the well cleanup efficiency. The second stage was the dynamic testing. In this, different flow conditions were considered to simulate the well production as the reservoir pressure depletes. Flow pattern, pressure gradient, liquid film reversal and liquid holdup distribution along the well were evaluated. Moreover, when slug flow was observed, slug length and frequency were calculated. When severe slugging was observed; slug cycle and maximum pressure at the bottom of the vertical section were calculated. Static experimental results indicated that lateral section configuration has a significant effect on the unloading efficiency under static conditions. Toe-up configuration requires lower gas flow rates to assure high unloading efficiency, and it is recommended as the optimal well trajectory to assure the well cleanup after the well completion and fracturing. Dynamic experimental results indicated that at high gas flow rate well trajectory does not affect well performance. Furthermore, at high liquid flow rates, well trajectory effects diminish. At low gas flow rates, erratic production conditions such as liquid loading, slugging and severe slugging are observed. Moreover, well configuration significantly affects the slug flow development along the lateral and vertical section. Liquid loading is one of the most frequently occurring erratic production conditions. As a consequence, a model to predict the liquid loading onset along the well trajectory was developed. For this, liquid loading onset (LLO) was defined based on two possible mechanisms, namely, liquid film reversal (Barnea 1986) and liquid wave growth (Taitel 1976). In addition, liquid entrainment in the gas core and the effect of deviation angle on the liquid distribution along the pipe circumference were considered. Model performance was validated comparing predicted critical gas velocity against available experimental and field data, previous liquid loading onset models prediction and steady-state simulations in OLGA. Finally, liquid loading onset detection software was developed to predict the critical gas velocity along the well trajectory and different times based on the expected production decline. This allows the determination of the time and location (when/where) liquid loading onset starts and gas flow rate required to avoid liquid accumulation. |
