Energy · Crude oil
Crude oil properties from one NMR scan, validated against your methods
For refinery, terminal and blending laboratories. Rombo AI Platform turns a benchtop NMR scan of a crude or blend into several property estimates, each benchmarked against your ASTM reference method with metrics agreed before a pilot. Your methods stay the reference, and your team keeps the decision.
Use case — Oil & Gas
A faster read on crudes and blends between full assays
Crude quality decisions rest on a chain of tests: density and API gravity, sulfur, acidity, distillation, and compatibility checks when cargoes are blended. Each test has its own method and instrument. Rombo AI Platform is designed to add a fast screening layer to that chain: one NMR scan of the sample, several property estimates, and a flag when a crude or blend moves away from what you expected.
Samples that look different can go to the full method sooner. Which properties qualify is decided by validation on your crudes, not by a datasheet. Our guide to crude oil analysis methods maps what refineries test, where samples are taken and what each result decides.
The customer problem
Refineries and terminals handle crude slates that change by supplier, cargo and field. The reference tests are well established. The difficulty is time and coverage: many properties, many methods, and the most complete studies are too costly to repeat for every cargo.
One property, one method
Density, sulfur, acidity and distillation each need their own instrument, procedure and trained analyst, so a broad panel on every cargo takes laboratory time.
Full assays stay periodic
Full assays, compatibility studies and tank profiling are too resource-intensive to repeat for every operating change, so some decisions rest on older or partial data.
Models that do not travel
Property models built for one feedstock or one analyzer often need new reference data when an unfamiliar crude arrives or the instrument changes. See why models break when instruments change.
The Rombo AI solution
Rombo AI delivers validated properties, not another model to build and maintain. Rombo AI Platform starts from an NMR foundation model pre-trained on a large library of spectra, then calibrates it on your crudes and reference results. It is designed to need fewer reference samples per property than a model trained from scratch, and to transfer across instruments and sites; both are checked during the pilot.
Several properties from one scan
NMR AI Analyzer acquires the benchtop NMR spectrum and returns estimates for the properties that passed validation in your pilot. API gravity, sulfur and TAN are typical candidates: see what changes for the crude lab.
Validated against your methods
Each estimate is benchmarked on paired samples against your ASTM reference method, with acceptance metrics agreed before the pilot starts.
A fingerprint for every crude
Compare each cargo or blend with its expected quality and with earlier deliveries, and flag samples that move away from their history. Drift monitoring is configured during the pilot.
Traceable results
Model versions and audit logs record which model produced each result, so every number can be traced back and re-checked against the reference method.
Main applications
What NMR does not replace
Yields and cuts still come from physical distillation such as ASTM D2892, and product release follows the methods in the specification. Blend stability is confirmed with dedicated tests such as ASTM D7157; we treat NMR-based compatibility screening as work in progress, to be validated on your crudes. A validated estimate supports routine decisions between reference tests; it does not certify a cargo. For a single, well-characterized stream monitored in line, an installed NIR analyzer can be the better fit: our comparison of lab testing, NIR and high-field NMR sets the options side by side.
From feasibility to production
- Feasibility review, free: we look at your crudes, reference data, sampling points and the decision you want to improve, and tell you whether a pilot makes sense.
- Pilot: a model calibrated on your crudes and reference results, with a validation report against your ASTM methods and the metrics agreed at the start. A limited set of paired samples is enough to test the first properties.
- Production: deployment on your instruments, with model versioning and audit logs. Integration with LIMS or plant systems is scoped after the pilot.
Only the properties that meet the agreed metrics move to production.
How it helps your team
- Crude and QC laboratory managers screen more cargoes and blends with one scan and send the ones that changed to the full method first.
- Process and planning engineers get more frequent property data on feeds and blends to compare options before execution. The decision stays with them.
- Analytical chemists and chemometricians keep the reference methods as the reference: the model is benchmarked against them, and transfer between instruments is tested, not assumed.
- Refinery leadership sees a measured pilot with acceptance criteria agreed up front, before any scale-up.