Formation pressure gradient prediction is important in drilling operations from technical and economical points of view. The pressure data while drilling can be obtained by pressure while drilling (PWD) tool which is costly and not available in most wells. The available correlations for pore pressure prediction depend on well logging, formation properties, and combination of logging and drilling parameters. These data are not available for all wells in all sections. The objective of this paper is to use artificial neural networks (ANNs) to develop a model to predict the formation pressure gradient in real-time using both mechanical and hydraulic drilling parameters data. The used parameters included rate of penetration (ROP), mud flow rate (Q), standpipe pressure (SPP), and rotary speed (RS). A dataset of around 3,100 field data points were utilized to provide the predictive model. A different set of data (92 points) unseen by the model was utilized for validating the proposed model. The model predicted the pressure gradient with a correlation coefficient (R) of 0.98 and average absolute percentage error (AAPE) of around 2%.
Formation pressure is the pressure exerted by the fluids within the rock pore space. The normal formation pressure at certain depth originates from the weight of the salt water column extended from the surface to the point of interest. The deviation from the normal trend can be described as abnormal which can be either subnormal or overpressure (Mouchet, J.P., and Mitchell, 1989). Normal pressure is not constant, and it depends on the amount of dissolved salts, fluid types, gas presences and temperature gradient. Commonly, normal pressure gradient is around 0.465 psi/ft (Rabia, 2002). Supernormal, also called overpressure or geopressured, is the pore pressure exceeding the normal hydrostatic pressure. This pressure is created from normal pressure in addition to an extra pressure source. The excess pressure may be attributed to different mechanical, geochemical, geothermal, geological and combined reasons (Rabia, 2002). Overpressure zones may lead to severe technical and economic issues such as kicks and blowouts. For optimum well control, it is important that not only the identification but also the magnitude of abnormal pressure should be known. Contrarily, subnormal pressure is the pore pressure lower than the normal pressure and may lead to loss of circulation and differential pipe sticking resulting in setting additional casing strings (higher drilling costs) (Rabia, 2002). Accurate real-time pore pressure prediction may provide enhanced well path and casing design, better wellbore stability analysis, effective mud program and reduced overall drilling time and cost (Tingay et al., 2009; Zoback, 2007).
