Prevention of Drilling Problems Using Real-Time Symptom Detection and Physical Models

Abstract

Drilling problems can be reduced and better drilling decisions can be made when they are based upon the result of engineering calculations performed in real-time while drilling. Today drilling teams today are flooded with a wealth of raw data collected both on the surface and down hole, and decisions based upon this data are often dependent on the experience and intuition of a seasoned driller or drilling team. These decisions can be improved if the drilling team was continuously provided with guidelines showing the ideal operating envelope for all of the recorded data - guidelines based on calibrated engineering calculations performed in real-time. Prior to drilling a well, sophisticated pre-drill engineering calculations are performed to determine the torque, drag, hydraulic pressure, and other variables that are expected during the well construction process. Roadmaps are produced that show how these variables are expected to change with depth during the well construction process. This predrill planning has tremendous value to ensure that the correct equipment is deployed to meet the drilling objectives and to create the initial drilling plan. This modeling is also used to ensure that the proposed well plan predicts that the wellbore pressure can safely stay within the geo-pressure window, and that buckling, hole cleaning, and wellbore conditioning are considered. However, the predictions from predrill planning have very little use as a guide while actually drilling the well. The pre-drill model is based upon a proposed well path that very often is different from the actual well path; and the model is based on not-so-well-known estimates of frictions, weights, interior pipe diameters, exterior pipe diameters, hole size, and mud rheology. The pre-drill model becomes obsolete very quickly as the actual conditions deviate from the original plan. Until recently, the capability did not exist to perform the required torque and drag, hydraulic, and thermodynamic modeling in real-time. The required computing power and algorithms did not exist to solve the finite difference equations in real-time nor was there the capability to automatically calibrate the model in real-time. These limitations were removed when the International Research Institute of Stavanger developed a single integrated engineering model that has the capability to be driven by the driller's actions and solved and calibrated in real-time. The model links the mechanical, thermodynamic, and hydraulic effects with each other and compensates for the not-so-well-known model parameters through calibration. While drilling, this model continuously calculates the expected range of values for the surface and downhole sensors including:Hydraulic pressure at all depths (Transient ECD, ESD, including effects of cuttings)Hook loadSurface torquePit levels (including transient effects, effect of cuttings and effect of fluid compressibility).Cuttings location in real-timeOther key outputs that describe an ideal problem-free environment based on the actual rig activities (e.g. Hookload and Torque roadmaps, SPP, cutting loading index). These expected range of values (virtual sensor readings) are used by the drilling team in two ways:To provide guidelines for checking the quality of the dataTo detect symptoms of deteriorating hole conditions that often lead to drilling problems. In is important to note that these calculations have a foundation in pure engineering principles that have been in use for decades, but that have now been uniquely refined and applied in real-time while drilling. These calculations allow the drilling team to perform exception-based monitoring, which is a generally accepted engineering technique to identify conditions that may lead to future problems. This technique consists of comparing measured values from sensors to expected values in search of deviations. In order to apply this technique specifically to drilling operations, the capability must exist to model the operation in real-time. Virtual sensor outputs are calculated every second during all well construction activities (e.g. drilling, tripping, circulating, sliding, coring, displacing mud) in real-time. This allows a direct comparison to be made between the virtual sensor readings from the model with the raw data coming from the sensors. Deviations between these values provide engineers with the earliest possible evidence of deteriorating hole conditions that may lead to stuck pipe, kicks, lost circulation or hole collapse. This evidence precedes raw data rig site observations by hours or days, while providing the engineering support needed to take action and avoid problems. A unique feature is the capability to calibrate the model in real-time. After calibration, a good match between measured and modeled data confirms that ideal drilling conditions exist. Deviations indicate that conditions are deteriorating. This analysis can be performed on most of the curves presented today in the real-time operating centers, including hookload, standpipe pressure, PWD, surface torque, pit levels, or other sensor readings recorded during well construction. These calculations also provide the best indication of sensor data quality. Deviations between the sensor outputs and the expected range of values can be interpreted as either an indication of deteriorating conditions or of poor quality data (poor accuracy). This analysis often is the first indication available for drifting or damaged sensors. Calibrated engineering models can also provide value in two additional areas. The calculations can be performed as an investigative tool after a well has been drilled in order to determine the root cause(s) of non-productive time (NPT) related to stuck pipe, formation fracturing, fluid influx, and hole collapse utilizing first principles transient models. The analysis can also be used to drive an adaptive drilling automation system that takes control of the drilling equipment in response to changing downhole conditions. We will show case studies showing how Engineering-While-Drilling can be used to detect early symptoms and prevent problems such as packoffs, swabbing, hole collapse, kicks, lost circulation and the use EWD for drilling automation.