Predicting Wellbore Diameter from LWD Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for accurately measuring wellbore diameter during oil and gas extraction are hindered by drilling difficulties, leading to potential well integrity issues and increased costs due to the need for separate wireline caliper measurements.
Innovation Solution
A method using machine-learning models to predict wireline caliper log data from logging-while-drilling data, allowing for the estimation of wellbore diameter without the need for additional measurements, thereby reducing costs and operational time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If wireline multi-arm mechanical caliper tools are run in the wellbore after drilling to measure hole size, then measurement precision is improved, but productivity deteriorates due to additional rig time and operational delays
Solution Approach 1:
The patent applies preliminary action by collecting logging-while-drilling data during the drilling process itself, before the drilling operation is completed. This allows the machine learning model to predict caliper measurements without requiring a separate post-drilling wireline run, thus maintaining measurement precision while eliminating the productivity loss associated with additional rig time
Solution Approach 2:
The patent uses copying by creating a predictive model that replicates the function of physical caliper tools. The machine learning model is trained on paired data (LWD logs and actual caliper measurements) to learn the relationship between drilling parameters and wellbore diameter, then uses this learned pattern to generate accurate predictions without requiring the physical presence of caliper measurement tools
2Measurement precision
If a solo wireline logging run is conducted to acquire caliper data, then measurement precision is improved, but loss of time increases due to additional rig time requirements
Solution Approach 1:
The patent applies merging by combining the functions of multiple logging operations into a single integrated process. Logging-while-drilling tools simultaneously collect both the drilling parameter data and the wellbore diameter measurements during one continuous operation, eliminating the need for separate wireline logging runs and thereby reducing time loss while maintaining data quality
Solution Approach 2:
The patent implements universality by designing a system where logging-while-drilling data serves multiple purposes. The same LWD measurements that monitor drilling performance are also used as input features for predicting caliper measurements, making the data collection process multi-functional and eliminating redundant operations
3Measurement precision
If separate wireline caliper measurements are performed, then measurement precision is improved, but device complexity increases due to additional measurement equipment and operations
Solution Approach 1:
The patent introduces an intermediary - a machine learning model - that bridges the gap between indirect LWD measurements and direct caliper measurements. Instead of adding more physical measurement devices, the system uses software-based prediction to translate readily available LWD data into accurate wellbore diameter estimates, reducing equipment complexity while maintaining precision
Data Source
AI summary
A method to predict wireline caliper log data from logging-while-drilling data which includes collecting logging-while-drilling logs and caliper logs from a plurality of wells. The caliper log contains at least one channel. The method further includes pre-processing the logging-while-drilling data, selecting a subset of logs within the logging-while-drilling data, and aggregating the channels of the caliper logs forming aggregate logs. The method further includes splitting the pre-processed logging-while-drilling data and aggregate logs into train, validation, and test sets, wherein the validation and test sets may be the same, selecting a machine-learned model and architecture, and training the machine-learned model to form predicted aggregate logs from the logging-while-drilling data using the training set. Additionally, the method consists of using the machine-learned model to predict the aggregate logs using pre-processed logging-while-drilling data.