Meana Petro

Study leads, subsurface and facilities discipline managers, bid teams

Engineering Consultancies

Your product is studies. Most of the study budget goes into finding and cleaning data rather than into engineering.

What is hard today

  • Six weeks of a twelve-week study spent digitising PVT reports and reconciling well names
  • Client archives delivered as a hard drive of unsorted scans
  • No way to reuse the data preparation from the last study for the same field
  • Deliverables that cannot be audited back to source when the client challenges a number

What we do about it

  • Document OCR and structured extraction that turns a client archive into a queryable dataset in days
  • A PVT, well test and production repository that persists between studies
  • Automated well log QC and synthesis, which is usually the first two weeks of any subsurface study
  • Every extracted number carries a citation to its source page, so deliverables are auditable

Where this usually starts

  • DF-01Foundational

    Field Document OCR Pipeline

    Turns decades of scanned reports, logs and maps into searchable text, including Persian and Arabic documents and handwritten annotations.

    Data FoundationAI
  • DF-02Foundational

    Structured Extraction from Reports

    Pulls the numbers out of the reports — depths, rates, pressures, volumes, dates — and puts them into a database with a link back to the page they came from.

    Data FoundationAIDomain
  • SR-13Foundational

    PVT Modeling & EOS Tuning

    Extracts fluid study data from decades of laboratory reports and tunes the equation of state against all of it, not against the one report someone had to hand.

    Subsurface & Reservoir IntelligenceDomainAI
  • SR-07Foundational

    Well Log QC & Missing Log Synthesis

    Finds the bad intervals in the log archive and reconstructs curves that were never run, so the field has a complete set to work from.

    Subsurface & Reservoir IntelligenceAI

Questions we are asked

Can we deploy this per project?
Yes. A project-scoped deployment on your own infrastructure, with the client’s data segregated and removable at project close, is a supported configuration.
How much tuning does extraction need for a new client?
For common document types, little. For a client-specific report format with no equivalent in the trained set, expect a labelled sample of fifty to two hundred documents before extraction is reliable.