• norsk
    • English
  • English 
    • norsk
    • English
  • Login
View Item 
  •   Home
  • Fakultet for ingeniørvitenskap (IV)
  • Institutt for geovitenskap og petroleum
  • View Item
  •   Home
  • Fakultet for ingeniørvitenskap (IV)
  • Institutt for geovitenskap og petroleum
  • View Item
JavaScript is disabled for your browser. Some features of this site may not work without it.

Automatic Lithology Prediction on Well Logs Using Kernel Density Estimation Trained By Analysis of Cuttings Data

Strømslid, Simen Kristoffer Steiro
Master thesis
View/Open
17339_FULLTEXT.pdf (Locked)
17339_COVER.pdf (Locked)
URI
http://hdl.handle.net/11250/2615093
Date
2017
Metadata
Show full item record
Collections
  • Institutt for geovitenskap og petroleum [2196]
Abstract
Lithology interpretation is a vital part of the process when drilling oil and gas wells in the petroleum industry. The ability to predict what problems and challenges can occur along the well path and which precautions have to be taken provides a safer and more efficient drilling process for all parties involved. This interpretation is usually done manually by looking at seismic logs shot before drilling, core samples, traditional well logs and data from Logging While Drilling tools.

The idea of automating this process is nothing new, and was proposed as early as in the 80 s, with a method based on a statistical approach on available data to make a database of parameters which could easily be accessed and used for comparison. The approach used in this thesis is one of automating the interpretation process and use it in real-time on available data to provide not only information about the current formation, but to use this data to make statistically founded predictions about the lithology in front of the drill bit.

This thesis is largely based on work and methods developed by A. Corina (Automatic Lithology Prediction from Well Logging Using Kernel Density Estimation), and seeks to improve her work by using new logs to improve accuracy in the method of identifying lithology based on historical data.

The method for doing so was by using cuttings logs instead of geological interpretation to determine the baseline for which to build the predictions on, which should result in a higher accuracy and a lower misclassification rate.

The results did show a lower accuracy and higher rate of misclassification, which suggests that more research and data management is necessary.

One of the reasons behind this higher rate of misclassification can be the interpretation of the cuttings logs: even though the cuttings should in principle have a higher accuracy when determining the lithology, oftentimes we have several different types of formation at the same depth in the cuttings logs. In sections where we can have 50% sandstone, 45% shale and 5% carbonates, the assumption that the dominant lithology (in this case sandstone) is the only lithology at that depth could certainly skew the gamma ray measurements of that section and be a cause of misclassification.
Publisher
NTNU

Contact Us | Send Feedback

Privacy policy
DSpace software copyright © 2002-2019  DuraSpace

Service from  Unit
 

 

Browse

ArchiveCommunities & CollectionsBy Issue DateAuthorsTitlesSubjectsDocument TypesJournalsThis CollectionBy Issue DateAuthorsTitlesSubjectsDocument TypesJournals

My Account

Login

Statistics

View Usage Statistics

Contact Us | Send Feedback

Privacy policy
DSpace software copyright © 2002-2019  DuraSpace

Service from  Unit