OIL & GAS / USE CASE

LAS Curve Prediction & QC at Scale

Automating well-log quality control and reconstructing missing curves across large LAS and DLIS archives.

Explore the use case
Well-log curve reconstruction showing a missing curve and its predicted continuation
01 / THE CHALLENGE

Incomplete and inconsistent well logs slow down subsurface analysis.

Large legacy well-log archives contain missing, misaligned and noisy data, limiting reliable petrophysical analysis and modeling.

Missing Curves

Incomplete GR, DT and NPHI curves affect well-log interpretation.

Inconsistent Headers

Metadata and curve naming differences create schema mismatches.

Data Anomalies

Null values, sharp spikes and broken log intervals complicate QC.

Manual QC Effort

File-by-file cleaning is time-consuming and can introduce inconsistent results.

02 / THE SOLUTION

A hybrid AI workflow for curve prediction and log quality control.

The approach combines machine-learning prediction, metadata validation and anomaly detection to automate LAS/DLIS quality checks and curve reconstruction.

01

ML-Based Curve Prediction

Random Forest and Gradient Boosting models reconstruct missing GR, DT and NPHI curves.

02

Metadata & Schema Validation

Regex checks and PPDM schema rules validate and standardize curve headers.

03

Anomaly Detection

Identify and flag null values, spikes and broken log intervals for quality review.

04

Batch QC Processing

Apply automated checks and normalization across large LAS and DLIS file collections.

03 / PROJECT DEPLOYMENT

From large log archives to standardized, QC-ready well data.

The workflow was applied across more than 100,000 LAS and DLIS logs from multiple fields, basin types and well vintages.

01

Input Data

Over 100,000 LAS and DLIS logs from multiple fields, basin types and well vintages.

02

Processing Scope

LAS 2.0/3.0 and DLIS formats, with batch automation and field-scale normalization.

03

Final Output

QC-checked logs with predicted curves and standardized headers, ready for downstream ingestion.

Example of well-log curve reconstruction, with a missing curve restored alongside the original log tracks
CURVE RECONSTRUCTION

Predicted and standardized well-log curves

Reconstructed curves and consistent log headers support downstream petrophysical and AI workflows.

04 / IMPLEMENTATION

Well-log QC workflow at a glance.

01

LAS / DLIS Input

Ingest well-log files from large legacy archives for batch processing.

02

AI QC & Prediction

Validate metadata, detect anomalies and predict missing GR, DT and NPHI curves.

03

Standardized Output

Generate QC-checked logs with predicted curves and standardized headers.

Delivered through LogiSense™, bringing curve reconstruction, log standardization and quality control together in a unified well-log data workflow.

05 / BUSINESS VALUE

Faster quality control and greater well-log data utility.

The AI-powered workflow reduced manual effort and accelerated the preparation of consistent datasets for downstream use.

95%

QC Automation

Automation achieved in data quality checks across large legacy log libraries.

<2 min

Faster Curve Cleaning

Log cleaning and validation time reduced from 30 minutes to less than 2 minutes per file.

AI-ready

Normalized Log Data

Standardized logs support field-wide analytics, machine learning and interpretation workflows.

Reported outcomes from this use case. Results may vary with data quality, file characteristics and processing requirements.

POWERED BY NERGYFLUX

LogiSense™

AI-powered well-log intelligence for curve reconstruction, data quality control and petrophysical workflows.

Explore LogiSense™

Have a well-log data challenge to solve?

Let's explore how AI can improve your well-log quality control and data readiness.

Discuss this use case