OPTIMUM DEPLOYMENT OF NON-CONVENTIONAL WELLS
Nonconventional wells (i.e., wells with an arbitrary trajectory or multiple branches) offer great potential for the recovery of petroleum resources. Wells of this type are underutilized
in practice, however, in part because it is difficult to optimize their deployment. In this dissertation, we focus on the reservoir engineering aspects of the optimum deployment of
nonconventional wells. The effects of uncertain geological and engineering parameters are included in this optimization. To maximize reservoir performance (recovery or net present
value), we optimize the number of producers and injectors, their types (e.g., vertical, horizontal or multilateral), locations and trajectories, as well as their control strategy via smart
We apply a genetic algorithm (GA) as our master engine for the optimization of well type, location and trajectory. This engine is accompanied by an artificial neural network (ANN) which acts as a proxy to the reservoir simulations (objective function evaluations), a hill climber, which searches the local neighborhood of the current solution, and a near wellbore upscaling, which allows the incorporation of near wellbore heterogeneity from detailed reservoir descriptions into coarse simulation models. In addition, we introduce an experimental design methodology (ED) to reduce the number of simulations required to quantify the effects of the multiple uncertain parameters during this optimization process.
Within this framework we can account for the control of the wells through a “reactive” control strategy. Using such a strategy, downhole control devices can open or close depending
on the fluids produced from different segments of the well.
We also developed an optimization tool based on a nonlinear conjugate gradient algorithm that enables decisions regarding the deployment of smart completion technology.
This tool is independent of the well type, location and trajectory optimization. It allows us
to implement a “defensive” control strategy; i.e., the control devices are opened or closed based on a well control optimization. With this strategy, reservoirs can be screened for
smart well technology. Reservoir uncertainty can also be accounted for within this framework.
We present single and multiple well deployment examples for different synthetic reservoir models. In these examples, well type, location and trajectory are optimized. The effects of uncertainty are included in several of the examples. We determine sensible single and multiple well deployment plans with the algorithms developed. We show that the objective function (cumulative oil produced or net present value of the project) is always increased relative to its value in the first generation of the optimization, in some cases by 30% or more. The optimal well type is found to vary depending on the reservoir model and objective function. We also show that the optimal type of well can differ depending on whether single or multiple realizations of the reservoir geology are considered.