This is a tough one. Unlike stick slip, the signs are way more subtle. This post will inform you what we have learned by looking at data. We hope you find it helpful.
To identify bit balling from surface data in real-time, we need to be looking for a long-term decrease in torque as well as long-term decrease in ROP. Understandably, that could be due to a variety of other reasons as well. We therefore track MSE (for a decreasing non erratic trend), depth of cut (again for a decreasing non erratic trend) to help support the hypothesis. We rule out stick slip and bit bounce to some extent using the Torque Variation Index and WOB Variation Index (more on this next week when we discuss bit bounce). Mud type is a major influence, and so is hole depth. Putting all this together we do get a pretty good bit balling detection model, which works for the most part. Contextual information that the rig crew generally have, that is currently not digitally captured, can for the for the most part eliminate false alerts.
The slides below are quite self explanatory. We invite you to download it, and study it. The proof is in the pudding they say. It has worked well for us so far. Give us the datasets you want to test this model on. We will be happy to do those tests for free. They also say all models are wrong, some are useful. Advise us on know how we can make this model more accurate and useful.
Note that the balled bit in the slide # 1 does not correspond to the data shown on the right. For those interested in further information, please refer to SPE-204063-MS, or contact us at firstname.lastname@example.org. Next week we will look into bit bounce detection, and soon (in three weeks time), we will show how all the models we have discussed so far – will help optimize ROP – which as you already know, is one of the most coveted prizes in well construction.
Click below for additional slides on this topic: