Prediction of Rate of Penetration in Drilling

Abstract

Drilling has become an expensive and necessary operation to explore petroleum and natural gases.The goal is to increase drilling speed with minimum cost and at the same time maintaining safety.Predicting rate of penetration (ROP) in real time is very important to optimise drilling cost , since the greater the ROP is, the lesser the drilling cost would be.In this work,the typical extreme learning machine (ELM) and an e cient learning model (Arti cial Neural Network) have been used in ROP prediction.Since the relationship between ROP and the set of parameters a ect ROP is highly non linear, these learning models have been used to capture the non linearity. The models have been built using WITSML Realtime drilling data which is open source data and hence erroneous.Results indicate that both ELM and ANN are competent for ROP prediction, though ELM has higher learning speed compared to ANN.Though the quality of the data set is not upto the mark, still we managed to achieve around 70% and 62% MAPE (Mean absolute percentage error) score for ELM and ANN model respectively.This work will help drilling engineers to predict ROP according to their computation and accuracy demand. Now we shall discuss about a di erent work on computer vision. Sewer pipeline networks have become the main concern of modern municipalities around the world as these networks are too old and they are reaching their design lifetime; meanwhile, increasing environmental and health requirements, growing demands, and tight budgets have all made the problem harder to deal with. In order to prevent severity and costly damage, sewer system con- ditions need to be monitored through a timely and comprehensive periodic assessment. Currently Manual inspection is common practice in the inspection and assessment of sewer networks. Visual inspection requires hundreds of hours of data processing by certi ed inspectors to detect defects (i.e., crack, joint o set, roots, deposit, in ltration, etc.) and assess defect severity (i.e., length, number, consequences, etc.). However, manual inspection used in the assessment of extensive sewer systems is error-prone, subjective, and time-consuming. And due to this sometimes-di erent type of defects in sewer pipe system go undetected until they do a good bit of damage. Here our objective is to automation model development using computer vision techniques for sewer condition assessment and automation of classi cation and detection of di erent type of defects (i.e., Crack, Fracture, Obstruction etc) which are found in sewer pipe system.