Analytical instrumentation
Petroleum fuel is among the top used energy sources in the world, with petroleum products making up roughly a third of total global energy consumption [1].
Petroleum laboratories are an essential resource for quality assurance and process optimization of such fuels, ensuring the purest product.
They allow verification that fuels follow industry standards and maintain consistency, building public trust in test results [2].
Traditional laboratories operate with workflows in which individual devices produce a report that must be manually reviewed.
This is extremely time-consuming and can cause inconsistencies in quality. Integration of artificial intelligence (AI) makes laboratory automation more efficient by focusing on data-centric workflow.
This includes automated processes and the streamlining of information systems into unified networks, ultimately leading to reductions in manual intervention [3].
Deep learning modules have been used in petroleum analysis, producing accurate predictions in fuel properties from spectroscopic data [4].
This method allows efficient organization of gasoline samples using advanced gas chromatography and mass spectrometry measurements [5].
Similarly, open-source AI software has been developed to automate gas chromatography–mass spectrometry (GC-MS) data analysis, showing results in reducing manual data handling [6].
Analytical results, such as chromatography outputs, spectroscopy measurements, and visual inspections, still require heavy manual intervention.
Machine learning is also able to take on this labor-inducing work and even produce results better than humans [7].
Petroleum laboratories also focus on core and rock research to analyze sediment samples taken from underground reservoirs via petrographic image analysis, which is essential in determining the amount and permeability of the oil and gas in a given reservoir.
However, the image-based measurements used to find this data are another strenuous task and prone to human error.
Fortunately, AI also allows these images to be analyzed automatically and more consistently [8].
All of this is evidence that AI can aid petroleum laboratories in doing more than just replicating existing test results; it can extend measurement capabilities by scouting out discrepancies earlier and stabilizing uncertainty.
This paper explores research and industrial developments published between 2022 and 2025 on how AI has been used to improve petroleum laboratory testing and instrumentation.
The focus is on data-driven techniques, ranging from spectroscopy prediction, automated chromatogram interpretation, image-based measurements in core and rock labs, and the digitalization of petroleum laboratory workflows.
Spectroscopy studies how matter interacts with light to determine chemical composition and material properties.
Petroleum laboratories frequently use spectroscopy, as it allows for efficient and nondestructive analysis of fuels.
Due to the simple and cost-effective nature of spectroscopy, it has been extremely attractive for AI usage to improve speed and interpretation.
In a 2025 study conducted by Haffner et al. [4], the research group developed “Inception for Petroleum Analysis” (IPA), an AI-based neural network that allows prediction of fuel properties from spectroscopic data.
A key property in petroleum testing is the cetane number, which measures the ignition quality of diesel and other middle distillate fuels [4].
Traditional methods require the cetane number to be estimated using chemometric models such as partial least squares (PLS), which rely heavily on manually designed features and struggle with more complex sets of data [4].
What sets IPA apart is that it allows analysis of relevant patterns directly from the spectra [4].
When applied to the prediction of cetane number for diesel and related fuels, the AI-based model achieved regression errors that were 40% lower than those of PLS [4].
As seen in Figure 1, Haffner was able to obtain the cetane number of 249 middle-distillates samples such as kerosene and diesel given ASTM D613-95 [4].
These results in Figure 1 show a smooth and wide distribution for the cetane property which is rather difficult for most conventional manual determination methods and even some AI models, indicating a high potential for IPA [4].
Similar ideas are also being explored using different forms of easily available data.
Other recent studies have focused on AI to estimate ignition-quality metrics from molecular descriptors or basic physicochemical properties.
For instance, the 2025 study conducted by Huggins et al. [9] created machine-learning models to predict the cetane number of sustainable aviation fuels using such inputs.
Their models achieved 5–10% prediction errors, which is like the results of standardized ignition quality test methods, sufficient for many screening and quality control applications while being much faster and more consistent in regular use [9].
In addition to predicting fuel properties such as the cetane number, AI-enabled infrared spectroscopy can be applied to classification tasks.
This application focuses more on identifying whether the fuel meets requirements or has been tampered with.
There are several studies that provide clear evidence of learning models being able to reliably pick out compliant gasoline samples.
One example is the 2024 study conducted by Biaktluanga et al. [10], where they used the infrared spectra and AI to evaluate gasoline quality from samples from three companies in India.
The model learned to distinguish between different quality levels of gasoline of whether they are suitable for fuel or not directly from the varying spectrum measurements, completely avoiding having to manually check chemical testing [10].
Similarly, the 2024 study conducted by Lalramnghaka et al. [11] applied infrared spectroscopy with data-driven models to detect and estimate the level of fuel adjusting in India.
They used spectral data as the input to automatically identify whether fuel had been tampered with, whether it was diluted with another chemical or contaminated [11].
Another version of spectroscopy is Raman spectroscopy, an optical technique that provides a molecular fingerprint of petroleum products.
This replaces the traditional technique used to identify the type of petroleum fluid, chromatography, a technique used to separate complex mixtures, which can be very time-consuming.
More recently, Raman spectroscopy has been used along with machine learning to enable faster identification.
For example, the 2025 study conducted by Hodges et al. [12] revealed that this method can accurately distinguish between synthetic and petroleum hydraulic fluids.
They trained the models to learn the spectral patterns of the two fluid types and then tested them on unseen samples.
AI-based models can point petroleum laboratories toward faster workflows, reserving more expensive confirmatory testing for samples that show higher uncertainty.
As datasets grow larger and models become more reliable, these tools will allow routine tests to move away from slow, fully manual procedures.
Another technique frequently used in petroleum laboratories is chromatography, which allows for separation of complex mixtures and analysis of individual components.
As it relates to petroleum, gas chromatography can separate and identify hydrocarbons present via mass spectroscopy or flame ionization to determine impurities, additives, and composition [13].
However, while chromatography is extremely useful, it can be time-consuming and expensive to conduct.
As such, many recent studies focus on two solutions: finding alternative options instead of chromatography or automation of results through data-driven models.
A 2022 study conducted by Alizadeh et al. [13] revealed that infrared spectroscopy combined with multivariate models can be used to estimate properties.
The infrared spectra were collected from crude oil samples and used as data input for multivariate regression models, which eventually learned how to distinguish poor samples from target properties [13].
Other studies, such as the previously mentioned studies by Biaktluanga et al. [10] and Lalramnghaka et al. [11], also demonstrated similar results.
These methods opt towards the first solution, removing the need for chromatographic separation by using spectral measurements with AI analysis instead for the same intended result.
There have been advancements towards the other method.
A 2022 study conducted by Gao et al. [14] used machine-learning models for prediction of viscosity of heavy oils diluted with lighter oils.
In their study, using trained data-driven artificial neural network models (Figure 2) with input variables related to the oil blends and operating conditions allowed the model to provide the necessary viscosity estimates without requiring extra manual work [14].
Similarly, a 2024 study conducted by Mourched et al. [15] used AI to detect diesel alteration using only easily measured processes and physical variables, as seen in Figure 3 via a sensor.
They collected routine measurement data present in existing process environments to train models to recognize patterns in diesel that have been tampered with and those that have not [15].
This process used to require manual chromatography.
There have also been instances of machine learning being applied directly to chromatographic data.
A 2025 study by Kaspi et al. [16], for example, used AI trained on gas chromatography spectra to identify petroleum distillates and gasoline. Identifying these substances is important in many petroleum applications, allowing insight into contamination investigations.
The models learned to recognize these patterns, reducing the need for expert interpretation, making the analysis much faster [16].
Figure 2: Schematic of the network in an Artificial Neural Network Model where the equation represents sum of inputs. Reproduced from Gao et al., (2022) [14].
Figure 3: Visual representation displaying the 3D design of the proposed sensor, highlighting the light path from the laser source towards the sensing zone. Bachar Mourched et al., (2024) [15].
These advances allow for a more flexible and efficient laboratory workflow in which the tedious task of conducting chromatography is limited. This shift allows for more cost-effective and consistent testing.
An essential component of petroleum laboratory operations includes the testing done at core and rock laboratories.
By analyzing reservoir rock samples using petrographic image analysis, determination of how fuel is stored is possible.
This is the study of rock thin-section images to identify features that reveal reservoir quality.
Petroleum laboratories usually need experts to manually interpret.
Recent advances in computer vision and machine learning allow for automated analysis.
A 2024 study conducted by Azzam et al. [8] used their deep learning model, Grainsight, for automated petrographic image analysis.
Their results, as seen in Figure 4, showed a significant reduction in analysis time and the removal of bias [8].
More recently, a 2025 study conducted by Fan et al. [17] also used a foundation model for rock thin-section images, called RolmAI. It can be adapted to different tasks such as mineral identification or image segmentation.
This kind of computer vision–based analysis is likely to become a standard part of rock and core laboratories. These tools will allow for precise, effective measurements that expand on traditional laboratory techniques.
Current laboratory workflows involving sample preparation and interpretation rely on individual review of results by analysts.
Modern digital laboratories aim to support faster and more consistent reporting and quality control.
One issue prevalent in laboratories is the difficulty of estimating important variables for petrol quality, including sulfur, nitrogen or oxygen content.
For example, measuring sulfur content tends to be slow and costly. Soft sensors with AI models are one of the proposed solutions.
These sensors can infer sulfur content in refinery processes, providing accurate, timely results [18].
In a practical setting, the 2025 study conducted by Ujević Andrijić et al. [18] used data-driven soft sensor models to infer sulfur content from related process measurements.
They measured ongoing plant data, which allowed the sensors to learn to use these signals to estimate sulfur concentrations [18].
Even though these signals do not directly reveal how much sulfur there is, they do point to the desulfurization reactions that determine the final sulfur level [18].
This allows sulfur levels and other similar high difficulty-to-track variables to be monitored continuously, expanding on the benefits of utilizing AI.
A proposed idea of digitalization of petroleum laboratories is the Laboratory Information Management System (LIMS).
It is considered the first step towards utilizing AI in a lab. LabHQ, a company that specializes in LIMS, discusses that smaller labs should first start off adopting a LIMS, as the jump to AI might be too excessive [19].
LabLynx, another LIMS-dedicated company, explains how LIMS can track the progression of samples in a testing workflow, an especially important feature when dealing with multiple analyses of the same fuel [20].
In addition, this collected data can automatically transfer into a centralized system to reduce transcription errors [20].
These developments demonstrate the push of digitization that allows for the removal of manual review.
The combination of AI tools and LIMS platforms transforms petroleum laboratories into places that emphasize efficiency and quality.
AI will continue to expand and integrate itself into petroleum laboratories, but there are barriers that limit worldwide spread.
One major issue is data availability.
Petroleum laboratory datasets are very site-specific and vary widely.
Model performance can drastically drop when conditions deviate even slightly from what the model has already observed.
One practical example is how Kaspi et al. [16] focused on a combination of measured chromatography spectra as well as external testing to confirm generalization.
Another issue is the trustworthiness of AI data. Ujević Andrijić et al. [19] showed that despite soft sensors being quite efficient, such systems still require careful monitoring to ensure that there are no discrepancies occurring.
These issues are summarized by Nam & Park [3] in Figure 5.
Trends point to multimodal learning as the next objective, where spectra, images, and operational metadata are fused into joint models.
There is also the long-term vision of automatic labs, where AI and robotics work in tandem to conduct experiments with minimal human supervision [21].
Over the past three years, artificial intelligence has evolved from a prospective concept to actual tools, heavily improving petroleum laboratory testing and instrumentation.
AI used in spectroscopy can outperform traditional methods with rapid estimates on key information.
The different solutions to resolve the tedious work of chromatography will also allow for faster and more cost-effective experiments.
Rock and core laboratories will also benefit from the AI spread, reducing manual interpretation.
Soft sensors and LIMS will carve the way towards true digitalization.
One essential aspect to consider is the idea that AI should not be completely replacing traditional laboratory methods.
By balancing the tools at hand, petroleum labs will likely expand into an operation with faster turnaround and improved consistency.
Dr. Raj Shah is a Director at Koehler Instrument Company in Holtsville, New York, where he has served for over three decades, contributing to the advancement of petroleum, fuels, lubricants, and analytical instrumentation technologies worldwide.
Over the course of his distinguished career in the energy and chemical engineering industries, he has become widely recognized for both his technical leadership and sustained service to global professional societies.
Dr. Shah is an elected Fellow by his peers at ASTM International, the Institute of Chemical Engineers (IChemE), the Chartered Management Institute (CMI), the Society of Tribologists and Lubrication Engineers (STLE), the American Institute of Chemists (AIC), the National Lubricating Grease Institute (NLGI), the Institute of Measurement and Control (InstMC), the American Oil Chemists’ Society (AOCS), the Institute of Physics (IOP), The Energy Institute (EI), and The Royal Society of Chemistry (RSC).
These fellowships reflect his multidisciplinary impact across chemical engineering, tribology, measurement science, energy technology, and applied chemistry.
He is also the recipient of the prestigious ASTM Eagle Award and ASTM’s highest honor, the Award of Merit (Fellow), recognizing more than 30 years of leadership and contribution to Committee D02 on Petroleum Products, Liquid Fuels, and Lubricants.
He recently co-edited the bestseller, Fuels and Lubricants Handbook: Technology, Performance, Properties, and Testing, a major reference work for the industry.
Dr. Shah has now authored and co-authored over 750 technical publications, conference papers, and industry articles, and continues to be an active contributor to the scientific and engineering literature.
Further information regarding his work and recognitions can be found at https://shorturl.at/I7000.
Dr. Shah earned his doctorate in Chemical Engineering from The Pennsylvania State University and is a Fellow of The Chartered Management Institute, London.
He is a Chartered Scientist (CSci) with the Science Council, a Chartered Chemist (CChem) with the Royal Society of Chemistry, a Chartered Engineer (CEng) with the Engineering Council, UK, and a Chartered Petroleum Engineer (CPEng) with the Energy Institute.
He was recently granted the honorific distinction of “Eminent Engineer” by Tau Beta Pi, the oldest and largest engineering honor society in the United States, an honor reserved for engineers demonstrating exceptional professional achievement and character.
Actively engaged in academia and mentorship, Dr. Shah serves on the Advisory Boards of Farmingdale State College (Mechanical Technology and Engineering Management), Auburn University (Tribology and Lubrication Science), and the State University of New York at Stony Brook (Chemical Engineering and Materials Science & Engineering).
He is also an Adjunct Professor in the Department of Materials Science and Chemical Engineering at Stony Brook University.
Throughout his career, he has remained deeply committed to advancing engineering education, standards development, and technical excellence within the global energy community.
More information on Dr. Shah can be found at https://shorturl.at/yYl85.
Dr. Vikram Mittal, PhD is an Associate Professor in the Department of Systems Engineering at the United States Military Academy.
His research interests include energy modeling, technology forecasting, and Alternative fuels.
Previously, he was a senior mechanical engineer at the Charles Stark Draper Laboratory. He holds a PhD in Mechanical Engineering from MIT, an MS in Engineering Sciences from Oxford, and a BS in Aeronautics from Caltech.
Dr. Mittal is also a combat veteran and a major in the U.S. Army Reserve.
Mr. Justin Zheng is part of a thriving internship program at Koehler Instrument Company in Holtsville, NY underneath Dr. Raj Shah.
Mr. Gavin Thomas is part of a thriving internship program at Koehler Instrument Company in Holtsville, NY and is a recent graduate of the Chemical and Molecular Engineering program at Stony Brook University.
He also works as a process engineer at Mill-Max in Oyster Bay, NY where he becomes hands-on with various production processes to ultimately improve safety, efficiency, and cost-effectiveness.
[1] EIA. (2022, February 25). Oil prices and outlook - U.S. Energy Information Administration (EIA). Eia.gov. https://www.eia.gov/energyexplained/oil-and-petroleum-products/prices-and-outlook.php
[2] Rand, S. (2012). Program Handbook for Engine Fuels, Petroleum and Lubricant Laboratories.
[3] Nam, Y., & Park, H.-D. (2025). Revolutionizing Laboratory Practices: Pioneering Trends in Total Laboratory Automation. Annals of Laboratory Medicine. https://doi.org/10.3343/alm.2024.0581
[4] Haffner, F., Lacoue-Negre, M., Pirayre, A., Gonçalves, D., Gornay, J., & Moreaud, M. (2025). IPA: A deep CNN based on Inception for Petroleum Analysis. Fuel, 379, 133016. https://doi.org/10.1016/j.fuel.2024.133016
[5] Nguyen, H. M., Sühring, R., Marx, C., Liang, Y., Sandau, C., & O’Sullivan, G. (2025).
An open-access computational fingerprinting workflow for source classifications of neat gasoline using GC × GC-TOFMS and Machine Learning. Journal of Chromatography A, 1762, 466388–466388. https://doi.org/10.1016/j.chroma.2025.466388
[6] Pecchi, M., & Goldfarb, J. L. (2024). Open-source Python module to automate GC-MS data analysis developed in the context of bio-oil analyses. RSC Sustainability, 2(5), 1444–1455. https://doi.org/10.1039/d3su00345k
[7] Babu, B. K., Murali Manohar Yadav, Singh, S., & Vijay Kumar Yadav. (2024). Fuel forensics: Recent advancements in profiling of adulterated fuels by ATR-FTIR spectroscopy and chemometric approaches.
Spectrochimica Acta. Part A, Molecular and Biomolecular Spectroscopy, 312, 124049–124049. https://doi.org/10.1016/j.saa.2024.124049
[8] Azzam, F., Blaise, T., & Brigaud, B. (2024). Automated petrographic image analysis by supervised and unsupervised machine learning methods. Sedimentologika, 2(2). https://doi.org/10.57035/journals/sdk.2024.e22.1594
[9] Huggins, P., Martin, J. S., Downey, A. R. J., & Won, S. H. (2024). Interpretable machine learning for predicting the derived cetane number of jet fuels using compact TD-NMR.
Sensors and Actuators B: Chemical, 426, 137018. https://doi.org/10.1016/j.snb.2024.137018
[10] Biaktluanga, L., Kharlukhi, L., & Choudhary, V. (2024). Analysis of gasoline quality by ATR‑FTIR spectroscopy with multivariate techniques. Results in Chemistry, 8, 101575. https://doi.org/10.1016/j.rechem.2024.101575
[11] Lalramnghaka, M., Lalfakawma, R., & colleagues. (2024). Detection and estimation of adulterated gasoline fuel in India using FTIR‑ATR spectroscopy and chemometrics. Infrared Physics & Technology, 105119. https://doi.org/10.1016/j.infrared.2024.105119
[12] Hodges, J. E., Roberts, A., & colleagues. (2025).
The classification of synthetic‑ and petroleum‑based hydraulic fluids using handheld Raman spectroscopy and machine learning. Chemosensors, 13(9), 327. https://doi.org/10.3390/chemosensors13090327
[13] Alizadeh, S., Ta, S., Ray, A. K., & Lakshminarayanan, S. (2022). Determination of density and viscosity of crude oil samples from FTIR data using multivariate regression, variable selection and classification. IFAC‑PapersOnLine, 55(7), 845–850. https://doi.org/10.1016/j.ifacol.2022.07.550
[14] Gao, X., Dong, P., Cui, J., & Gao, Q. (2022). Prediction Model for the Viscosity of Heavy Oil Diluted with Light Oil Using Machine Learning Techniques. Energies, 15(6), 2297. https://doi.org/10.3390/en15062297
[15] Bachar Mourched, AlZoubi, T., & Sabahudin Vrtagic. (2024). Diesel Adulteration Detection with a Machine Learning-Enhanced Laser Sensor Approach. Processes, 12(4), 798–798. https://doi.org/10.3390/pr12040798
[16] Kaspi, O., Avissar, Y. Y., Grafit, A., Chibel, R., Girshevitz, O., & Senderowitz, H. (2025). Machine Learning-Based Identification of Petroleum Distillates and Gasoline Traces Using Measured and Synthetic GC Spectra from Collected Samples. Molecular Informatics, 44(8), e202400371. https://doi.org/10.1002/minf.70008
[17] Fan, J., Yu, X., Di, Y., Tianxu Lv, Zhang, R., Bao, J., Liu, Y., Li, L., & Pan, X. (2025). A foundation model for rock thin-section images analysis. Communications Engineering, 5(1), 9–9. https://doi.org/10.1038/s44172-025-00565-5
[18] Ujević Andrijić, Ž., & colleagues. (2025). Intelligent soft sensors for inferential monitoring of sulfur content in hydrodesulfurization and related petroleum processes. Actuators, 14(8), 410. https://doi.org/10.3390/act14080410
[19] Digitization in Oil and Gas: Practical Technology Solutions Driving Efficiency and Quality. (2026). Thelabhq.com. https://www.thelabhq.com/resource-centre/digitization-in-oil-and-gas-practical-technology-solutions-driving-efficiency-and-quality
[20] Louderback, T. (2023, January 25). The Role of Labs & LIMS in Oil & Gas Exploration. LabLynx - Advanced LIMS and ELN Solutions for Smarter Laboratory Management. https://www.lablynx.com/resources/articles/oil-and-gas-exploration-the-role-of-the-lab-and-lims-software/
[21] Sircar, A., Yadav, K., Rayavarapu, K., Bist, N., & Oza, H. (2021). Application of machine learning and artificial intelligence in oil and gas industry. Petroleum Research, 6(4). https://doi.org/10.1016/j.ptlrs.2021.05.009
PIN 27.2 Apr/May 2026
.jpg)
Product directory
