Real-time drilling optimization is a topic of significant interest because of its economic value, and its importance increases particularly during periods of low oil prices. This paper evaluates different optimization strategies and algorithms for real-time optimization of an objective function (function to be optimized) specific to drilling. The objective function optimized here is derived from a data-driven (or machine-learning) model with an unknown functional form. A data-driven model has been used to calculate the objective function [rate of penetration (ROP)] because it has been shown to be more efficient in ROP prediction relative to deterministic models (Hegde and Gray 2017). The data-driven ROP model is built using machine-learning algorithms; measured drilling parameters [weight on bit (WOB), revolutions per minute (rev/min), strength of rock, and flow rate] are used as inputs to predict the ROP.
Real-time drilling optimization that is data-driven is challenging because of run-time constraints. This is perceived as a handicap for data-driven models because their functional form is unknown, making them more difficult to optimize. This paper evaluates algorithms depending on their ability to best maximize the objective (ROP) and their time effectiveness. Two simple yet robust algorithms, the eyeball method and the random-search method, are presented as plausible solutions to this problem. These methods are then compared with popular metaheuristic algorithms, evaluating the tradeoff between improvement in the objective (search for a global optimal) and the computational time of run.
Using results from the simulations conducted in this paper, we concluded that data-driven models can be used for real-time drilling despite their computational constraints by choosing the right optimization algorithm. The best tradeoff in terms of ROP increase as well as computational efficiency evaluated in this paper is the simplex algorithm. The ROP was improved by 30% on average with a variance of 2.5% in the test set over 14 formations that were tested.
The matrix permeability of shale is controlled by the microstructure of shale pore network. Therefore, a thorough understanding of shale pore structure is fundamental to the prediction of shale permeability. We constructed a physically representative pore network model for two Barnett Shale samples. We predicted a Darcy permeability of 7.19 nanodarcies (nd) and apparent permeability of 57.34 nd for these samples. We explored the pore structure of shale matrix with low-pressure nitrogen adsorption/desorption isotherms. The pore size distribution, the network connectivity, as well as pore spatial arrangement are determined for the organic matter and the whole matrix. We separated pores in the network model into two groups: the affiliated pores and the dominant pores. For these samples, the affiliated pores are pores with diameters smaller than 8 nm and the dominant pores are with diameters larger than 8 nm. The affiliated pores develop on the walls of the dominant pores. The cutoff value between affiliated pores and dominant pores may differ for different samples. The pore spatial arrangement is validated against previous literature resultsand a diagenetic explanation for development of shale pore system. The network model can be used to predict shale permeability and other petrophysical properties.
Microfractures are important mechanical discontinuities in shales and are important for fluid flow during production. Understanding their properties is crucial for accurate shale production prediction and implementing effective stimulation strategies. Scanning electron microscope (SEM) images are useful for characterizing shale microstructure, but manual image analysis is often challenging and time consuming. We present an alternative method for quickly characterizing microfractures and obtaining pore structure information from SEM images using machine learning algorithms and image processing. Using this approach, SEM images were obtained from deformed and intact samples of a carbonate rich shale and a siliceous shale with the goal of identifying microfractures. Support vector machine, convolutional neural networks, and four pretrained convolutional neural networks were implemented to differentiate SEM images containing fractures (frac-images) and SEM images containing no fractures (non-frac-images). Images containing fractures were identified with 92% training accuracy and 88% testing accuracy. A pretrained convolutional neural network with 16 layers (vgg16) was shown to perform best for this image classification task.
Reduced-order models of the airwake produced by the flow over a simple frigate shipare developed using POD based methods. The focus is to understand the trade spacebetween cost and accuracy, where different forms of the POD technique are concerned.Of particular importance is the upfront expense of employing ‘classical’ or snapshot formsof the POD technique in both scalar and vector forms using either time suppressed data(conventional-POD), or kernels constructed from cross-spectral densities of the fluctuatingvelocity. The latter approach is referred to as harmonic-POD so as not to exclude harmonictransforms in space. The flow over a simple frigate ship is an ideal test bed given that it isunsteady, three-dimensional, inhomogenous in all spatial directions, and stationary in time. The spatial modes from all three techniques are shown to correspond to unique time-scales, thereby demonstrating how the preservation of the temporal behavior associated with a particular spatial scale is not unique to the harmonic-POD approach.
Water-in-water (w/w) emulsions are known for their low interfacial tensions (IFT) which makes their stability to shear questionable. This is because of low particle attachment energies, which can be just a few kT. Therefore, emulsions stabilized with larger particles should display greater stability to shear because of larger attachment energies (10–100 or more kT). This is typically not an issue with traditional oil-in-water Pickering emulsions because particle attachment energies are much larger due to higher interfacial tensions, even when very small particles are used.
Single-phase permeability k has intensively been investigated over the past several decades by means of experiments, theories and simulations. Although the effect of surface roughness on fluid flow and permeability in single pores and fractures as well as networks of fractures was studied previously, its influence on permeability in a random mass fractal porous medium constructed of pores of different sizes remained as an open question. In this study, we, therefore, address the effect of pore–solid interface roughness on single-phase flow in random fractal porous media. For this purpose, we apply a mass fractal model to construct porous media with a priori known mass fractal dimensions 2.579≤Dm≤2.893 characterizing both solid matrix and pore space. The pore–solid interface of the media is accordingly roughened using the Weierstrass–Mandelbrot approach and two parameters, i.e., surface fractal dimension Ds and root-mean-square (rms) roughness height. A single-relaxation-time lattice Boltzmann method is applied to simulate single-phase permeability in the corresponding porous media. Results indicate that permeability decreases sharply with increasing Ds from 1 to 1.1 regardless of Dm value, while k may slightly increase or decrease, depending on Dm, as Ds increases from 1.1 to 1.6.