Today applications of drilling require proper identification of operations where a cost reduction is possible. Many indicators are present when one tries to optimize the drilling operations such as casing size and mud properties. On the other hand the selection of the optimum bit requires information from a variety of sources. The parameters affecting the bit performance are complex and their relationship is not easily recognized. The general trend is to evaluate the performance of the bit from an offset well.
A new methodology was developed to model the rate of penetration and bit wear under various formation types and operating parameters. This method introduces a new approach with improved bit wear prediction. A simulator was used to generate drilling data to eliminate errors coherent to field measurements. The data generated was used to establish the relationship between the complex patterns such as weight on bit, rotary speed, pump rates, formation hardness, and bit type.
The method was tested using data from runs conducted with a rig floor simulator. The validity of the proposed method was also demonstrated with data from an existing field.
The success and hence the economics of a drilling operation depends on the condition of the bit. With bits performing at high penetration rates, the well drilling costs can be lowered. Thus, the selection of a proper bit type and the operating parameters are important challenges one faces during the drilling operations. Work performed by several investigators have shown that many bit and fluid components affect the penetration rates. Different methods can be utilized in the optimization of drilling. Researchers proposed the use of empirical correlations and predictive techniques. In these approaches either laboratory data were used to derive the empirical correlations or offset well data was used to fine tune the predictive method.
Neural Networks. Neural networks have been successfully used in different fields due to their capability to identify complex relationships when sufficient data exist. Recently, they have been successfully applied to different areas of petroleum engineering such as multi-phase pipe flow, reservoir characterization, production, and drill bit diagnostics. The neural network developed to diagnose the drill bit used six parameters consisting of lithology (or formation type), torque, rate of penetration, weight on bit, rotational speed, and hydraulic horsepower per square inch of nozzle as input. The network was trained to predict the bit wear as output. The use of formation type or lithology introduces errors for conditions where the predicted formation types and depths differ from the predicted properties. Although the drill bit diagnosis network was successful it was based on laboratory data and did not cover all formation hardness and bit grade levels, thus limiting its applicability.
In this study, we introduce a new approach to predict a drilling parameter such as the rate of penetration by designing a new neural network.
A new methodology is introduced to predict the ROP values during drilling. This approach uses the measured data to determine the relationship between several parameters like bit type, weight on bit, depth, and rotary speed recorded during the drilling operations.
Two different data sets were used in this study. The first data set consisted of approximately 8,000 measurements taken at selected wellbore conditions. The rig floor simulator available in the departmental facilities were employed for this purpose. The use of simulated data provided additional insight in terms of parameters like formation abrasiveness, bit tooth wear, and bit bearing wear as a function of drilling time that are commonly not possible to measure in the field. The second data set consisted of approximately 500 measurements from several wells in the United States. Simulated Data. Runs were conducted using a rig floor simulator and data were continuously recorded until bit fails either due to bearing wear or tooth wear. The data set contained approximately 8,000 measurements taken at predesigned wellbore conditions. The simulated data were chosen in this study to eliminate errors inherent to data acquired in the field. Table 1 shows the recorded data types and their range.