Real time pattern recognition in drilling logs

In short, the number of measurements in real time drilling logs is huge, but the measurements are often indirect for what is of real interest. Important features include for instance, the precise location of the hole, what formations are penetrated, what is the pore pressures in the formations, is the hole everywhere stable and clean and is the drill bit unworn and effective. Knowledge of these and other factors are important for optimizing the drilling process so that efficient drilling and low non-productive time is possible.
Three potential event types are proposed in this document. It is explained how they are linked to reducing the non-productive time during drilling. It is important to mention that the event types are potential and that the AMIDST project is not limited to detecting only these event types. It reflects the current knowledge of the tasks at hand.
1. Detection of underlying reasons for spikes in the pressure that is measured at the rig
When drilling a well, the drill bit is attached to a hollow pipe.  Drilling mud is for which is pumped from the rig and into this pipe. At the drill bit the mud is flushed through thin valves, which cleans the bottom of the hole and circulates back to the top on the outside of the hollow pipe.
The mud circulating system is in place for several reasons. One is to transport cuttings to the top and another is to balance the pressure in the wellbore (if the mud is taken out, the wellbore will surely collapse). A third reason is to drive a turbine in the mudmotor, to enable the drill bit to turn (or boost the rotation per minute, if the drill string is turned as well). A pressure measurement at the top is a measure for the overall hydraulic friction to drive the mud through the drill string, the turbine of the mudmotor, the valves at the bit and up through the wellbore. Obviously the pressure is a function of the flow rate, but it is also dependent on numerous other factors. It is of particular interest to detect when pressure spikes are a consequence of the mud motor stalling or when the hole is packing off. If the motor is stalling, the crew needs to deal with the equipment failure. In case of a packoff, the hole cleaning problem needs to be dealt with. There are different strategies to reduce the non-productive time if the reason for the pressure spike is known.

2. Detection of reasons for changes in pressure measurements at the bit
Analog to the pressure measurement at the top, this measurement is a measure of the hydraulic friction in the wellbore in addition to taking into account the density of cuttings in the wellbore and influx of formation fluids. Understanding the fluctuations in this signal in combination with other signal sources are key for identifying hole cleaning issues. Moreover, this signal can be an indicator of a hole stability problem, because influx of formation fluids or loss of mud into fractures will also increase this measurement.

3. Detection of formation type
Seismic images are used as background for specifying the three dimensional trajectory. Once the trajectory is given, a one-dimensional lithology chart can be derived, which is given before drilling a well. As the well is drilled, more information is available and updating a probabilistic model for dynamically updating the lithology chart is of interest. Other indications of hole cleaning or hole stability may be more or less severe, depending on the actual formation being penetrated.

The AMIDST impact
Using the AMIDST solution, it is expected that a subset of the events captured by DrillEdge is expected to improve, and possibly some new events will be possible as well. Moreover, the open source solution and the expected publications will add significant value to the drilling industry itself.

This website uses cookies. Cookies are only used for handling login and identifying registrered users, not for visitor tracking.