New Approach to Quantify Productivity of Fishbone Multilateral Well

Abstract

Abstract The fishbone wells are new production technology applied to increase well productivity and access the difficult geological formations and unconventional reservoirs. The main advantages of this technology over hydraulic fracturing are the competitive price and reduced operation time. Fishbone shaped multilateral wells proved better productivity than multi-fractured horizontal wells in relatively low permeable reservoirs. In this paper, a new approach is proposed to predict the fishbone performance without using smart well completion or any down-hole valves in the horizontal laterals. Very limited work has been done for multilateral fishbone drilled in dry gas reservoir, and few empirical models were developed to estimate the inflow performance of fishbone wells producing from two-phase reservoirs, however, those models ignore the number of rib holes and assume constant pressure drop across horizontal laterals, therefore, using such models can produce severe estimations errors. The main objective of this research is to present a reliable model to estimate the productivity of fishbone multilateral wells producing from anisotropic and heterogeneous gas reservoirs. Several artificial intelligence techniques were studied to quantify the productivity of fishbone multilateral well for a wider range of conditions without introducing uncertainties/ complexity associated with other numerical methods. The proposed models investigate the significance of reservoir parameters, number of laterals, permeability ratio (Kh/Kv), length of laterals and lateral spacing on the productivity of the fishbone well. More than 250 data sets were utilized to develop and validate the model reliability. The production rate is estimated using artificial neural network (ANN), adaptive neuro fuzzy inference system (ANFIS), generalized neural network (GRNN) and radial basis function network (RBF). The models require the reservoir parameters and the wellbore configurations to determine the flow rate without a need for using down-hole well completion. Furthermore, mathematical equation was extracted by utilizing the artificial neural network model, this equation was verified using two rate tests from actual gas field, an acceptable error of 7% was obtained. The finding of this work would afford an effective tool for quantifying the productivity of complex fishbone wells and refine the commercial well performance software to narrow down the differences between the simulation outputs and actual field data, then lead to a better determination of the optimum production rate.