Finding the Optimal Horizontal Well Trajectory using Monte Carlo Techniques: Implementation Details and Case Study in Abu Dhabi, UAE

Abstract

Optimal field development plans are often required to maximize reservoir recovery while keeping costs low. This paper discuss the details of how to select the optimal horizontal well trajectory that maximizes reservoir recovery for an interbedded thinly layered carbonate reservoir. The case study here was applied to one of the largest wet gas / gas condensate fields in the UAE. The development plan targets four different reservoirs using horizontal wells. Each reservoir has different rock qualities, and the top reservoir is not in communication with the three reservoirs below. Each reservoir contains 3-4 sub-layers with varying reservoir properties. Some of the sub-layers may not be in communication with the others, and the vertical communication could be poor. In order to maximize recovery, the development plan calls for placing the horizontal wells crossing from one sub-layer to another sub-layer. The problem is in deciding how long the horizontal well should stay in each sub-layer. Since there are four reservoirs and an average of three sub-layers per reservoir, there are twelve possible lateral placement options that control the well trajectory and length. The methodology presented in this paper utilizes Monte Carlo sampling to calculate the well trajectory that maximizes recovery. The methodology resembles the ideas of the Metropolis Algorithm used in the Marko Chain Monte Carlo. The starting point uses an arbitrary well trajectory. This well trajectory is run in the simulator, and the cumulative production is saved as a reference. Subsequently, the same distribution is sampled and the simulator is run again. If the resulting recovery is greater than the initial recovery or the recovery from any prior iteration step, then the newly found well trajectory is used as the new mean of the distribution, and the steps are repeated until simulated field recovery does not substantially increase.