Application of Artificial Neural Network and Hareland Model for Real-Time Prediction of Uniaxial Compressive Strength Using Mud Logging Data

Uniaxial Compressive Strength (UCS) of the rock is one of the key parameters in optimization of drilling parameters such as weight on bit (WOB) and rotary speed (RPM), Analysis of wellbore stability, planning mud weight, bit selection and determining bit wear. There are methods for prediction of UCS such as uniaxial compressive strength test, Schmidt hammer test, identation test and scratch test. These methods are most accurate methods but can’t provide a continuous rock strength profile along the wellbore. On the other hand, these methods need cores and coring is expensive and time consuming. Some researchers proposed correlations between UCS and formation properties using cores and logs. In previous researches, artificial neural network and adaptive neuro fuzzy inference system utilized for prediction of UCS using formation properties like porosity and density and logs like sonic log and density log. These methods can’t be utilized as a real-time method for prediction of UCS. In this study two methods are investigated for real-time prediction of UCS using mud logging unit data like WOB, RPM and rate of penetration. Hareland model and artificial neural network are utilized for prediction of UCS based on mud logging data. The models are constructed using offset wells data. Then models are tested for a new well. Results show the proficiency, acceptable accuracy and cheapness of these new methods.

کلیدواژه ها:

Unconfined compressive strength of the rock ، real-time ، Hareland model ، Artificial Neural Network ، drilling parameters