Share
  • When we drill fast (aiming for the best ROP possible), we also have to keep up with better hole cleaning actions. Otherwise, the time gained by faster drilling may very well be lost through subsequent potential stuck pipe events, loss in production, etc. While everyone understand that hole cleaning is important, automated hole condition monitoring tools are still not widely used. Cutting transport models are available and great for this purpose but running them in real-time is onerous due to the inputs these models require. The same is the case with the cuttings and cavings sensors available today. We, as an industry continue to improve them, but in the meantime, are there solutions we can use today to automatically track whether the hole is clean enough? The Bayesian network model presented in this post and deployed on rigs for over three years now, presents one such solution.

    The key idea here is threefold: track bad hole cleaning symptoms that can be seen in real-time data, understand the drilling circumstances, and track actions that the crew is taking that will lead to a cleaner hole. Drilling circumstances include factors such as hole angle, bit hydraulics, drilling circulation rate, activity in the hole, etc., which are in general assumed to be not easily controllable. Tight spots are potential symptoms of bad hole cleaning and are tracked for the purpose of understanding the possibility that cuttings are not getting removed. In addition, actions taken by the driller that promote cleaner hole such as working/rotating the pipe, and circulation when not drilling are also tracked. The model shown in Slide # 1 brings all this together.

    Slide 3 and 4 provide examples of how some of the individual nodes affect the hole cleaning belief. In Slide 3, it can be seen how working the pipe helps with the hole cleaning belief. The effect of a decreasing ROP can also be seen in this slide. Slide 4 shows how tight spots and a static hole contribute to lowering the hole cleaning belief. These examples are covered in more detail in SPE-204125 for those interested.

    The overall goal with this model is to alert the drilling crew that the hole condition may be bad, and that some action may be needed. The alert threshold should be set by the operator balancing the expected cost of a bad hole versus the time taken to clean a hole to “required quality” specification. We will go into a more detailed discussion on tight spots and the use of torque and drag models next week. So, stay tuned.

    Click below for slides on this topic:

    Hole Cleaning Monitoring