| Team Name |
Well Learners |
| Track |
Industry - Energy |
| Use Case |
Predict the zone in a well path |
| Technology |
SAS Viya, Python |
| Region |
EMEA |
| Team lead |
Nithya Mohan |
| Team members |
Looking for team members |
Predict zone for a well based on different well logs acquired along the reservoir part of wells. Use different Model types and benchmark which ones perform better. Currently zone prediction is done manually and sequentially one well at the time as wells are drilled in an area. The manual interpretation is done by Petro physicists using logs signature and general area and geological knowledge in addition to his/her knowledge and experience with the physical principles behind the log measurements. There is a potential to produce more consistent and unbiased predictions by data-driven methods.
A zone is a subdivision of the rocks in the subsurface based on rock quality and fluid properties, i.e., how porous is the rock, is it fill with water, oil or gas or a mix? is the rock dense and tight or porous and fractured etc., this leads to different log responses when it comes to sound velocity, nuclear background radiation etc. Note that only have indirect log measurement of the variation of physical properties along a borehole filled with drilling mud of a specific composition. We have no direct measurement of porosity or rock types. The problem is to predict a discrete zone (target) log from N (N<10 typically) feature (petrophysical and continuous) logs. Zone log is also often called lithology log.
I have a dataset with 28 030 011 rows and 136 columns.. I need to mask the well name, zone name and XY coordinates and then it’ll be ready for the hackathon by next week
