Real Time Prediction of Specific Energy During Laser Perforation in Limestone: A Neural Network Approach

摘要

In oil and gas well completion, perforation channels must be made through the steel casing wall and cement and into the rock formation in the production zone to allow formation fluid to enter the well. By developing the technology of high power lasers in recent years, applying high power lasers in perforating oil and gas wells will be so advantageous due to its economical and technical priorities over the explosive shaped charges. Unlike the conventional explosive shaped charge perforation that often causes great reduction of rock permeability, laser perforation would enhance the rock permeability which leads to increasing oil or gas production rate of the well. In this way, the efficiency of laser perforation can be determined due to specific energy. Specific energy is defined as the amount of energy required to remove a unit volume of rock. In this study, feed-forward with back-propagation and generalized regression neural networks have been designed to predict the specific energy during laser perforation in limestone which is one of the most common rock formations in oil and gas reservoirs. Effective parameters in laser perforation like laser power, lasing time, pulsation and pressure which are related to laboratory tests done by ytterbium-doped multi-clad fibre laser on core samples are the inputs and the specific energy is the output of the neural networks. The designed neural networks showed high correlation coefficients with low error and the specific energy for limestone was predicted successfully.