LAS Curve Prediction
& QC at Scale
Use Case
Scalable Quality Control and Curve Reconstruction for Well Logs
Business Challenge
Legacy well log archives contain thousands of LAS and DLIS files with missing, inconsistent, or noisy data. This severely limits subsurface analysis and petrophysical modeling.
- Vast libraries of well logs contained missing, misaligned, or incomplete curves, disrupting petrophysical analysis.
- Inconsistent metadata and header naming caused schema mismatches, complicating curve standardization.
- Frequent occurrences of null values, sharp spikes, and broken data segments led to significant quality control bottlenecks.
- Manual efforts for log cleaning were not only time-intensive but also introduced variability and errors due to human subjectivity.
The AI Approach
To overcome these issues, the client deployed a hybrid AI pipeline combining ML based curve prediction, schema validation, and anomaly detection to automate LAS/DLIS log quality control at scale.
- ML-based curve prediction was done using ensemble models like Random Forest and Gradient Boosting to reconstruct missing GR, DT, and NPHI logs.
- Metadata validation involved applying regex and PPDM schema rules to check and standardize curve headers.
- Anomaly detection was used to identify and flag nulls, spikes, and log breaks for QC automation.
- Batch processing support enabled scalable QC across thousands of LAS/DLIS files using a streamlined AI workflow.
Project Deployment Overview
Input Data Used
Over 100,000 LAS and DLIS logs sourced across multiple fields, basin types, and well vintages.
Final Output Generated
QC’d logs with predicted curves and standardized headers ready for ingestion into petrophysical and AI platforms.
Deployment Platform
Solution deployed via Python-based APIs and Streamlit front-end interface for interactive QC and review.
Processing Scope
Handled LAS 2.0/3.0 and DLIS formats, supporting batch automation and field-scale normalization.
Business Outcomes & Value Unlocked
The AI-powered log analytics workflow streamlined data quality control and curve reconstruction, reducing manual effort, accelerating log readiness, and unlocking consistent, trusted datasets for petrophysical modeling, machine learning, and basin-wide interpretation.
QC Automation at Scale
Achieved 95% automation in data quality checks across large-scale legacy log libraries.
Faster Curve Cleaning
Reduced log cleaning and validation time from 30 minutes to less than 2 minutes per file.
Normalized Logs for AI
Standardized logs enabled field-wide analytics, machine learning, and interpretation pipelines.
Maximized Data Utility
Transformed legacy logs into usable assets for modern modeling and multi-well correlation workflows.