Machine learning approach for predicting crude oil stability based on NMR spectroscopy

Raljević, Dubravka; Vuković, Jelena Parlov; Smrečki, Vilko; Pajc, Ljiljana Marinić; Novak, Predrag; Hrenar, Tomica; Jednačak, Tomislav; Konjević, Lucija; Pinević, Bruno; Gašparac, Tonka

doi:10.1016/j.fuel.2021.121561

Cited by 11 publications

(4 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There is no strict boundary between large resin molecules and asphaltene molecules, and in fact, they have very similar structural characteristics, namely one or more largesize FAR sheets. When the interactions between large resin molecules themselves, mainly through π-π stacking and surrounding π-σ interactions, are significantly stronger than their interactions with small resin molecules or polar SAoil molecules, they will aggregate and form structurally stable aggregates, [32,33] resulting in the diffusion portion centered around D r,HC-3 .…”

Section: Resultsmentioning

confidence: 99%

Diffusion‐Ordered Spectroscopy (DOSY) as a Powerful NMR Method to Investigate Molecular Mobilities and Intermolecular Interactions in Complex Oil Mixtures

Yuan,

Xu,

Yang

et al. 2023

ChemPhysChem

View full text Add to dashboard Cite

Studying the characteristics and molecular mechanisms of liquid self‐diffusion coefficient and viscosity changes is of great significance for, e.g., chemical and petroleum processing. As an example of a highly complex liquid, asphaltene‐free high‐acid and high‐viscosity crude oil and its extracted fractions were studied by comparing their 1H DOSY diffusion maps. The crude oil exhibited a polydisperse diffusion distribution, including multiple diffusion portions with diffusion coefficients much smaller than that of any single fraction in independent diffusion. The main mechanism that leads to the decreases in the diffusion coefficients of crude oil is attributed to diffusion resistance enhanced by Dynamical Molecular‐Interaction Networks (DMINs), rather than by enlargement of the diffusion species caused by molecular aggregation. Constructed through the synergistic interactions of various polar molecules in crude oil, DMINs dynamically bind polar molecules, trap polarizable molecules, and spatially hinder the free motion of non‐polar molecules. Overall, this reduces the diffusion coefficients of all molecular. This study demonstrates that DOSY is a powerful NMR method to investigate molecular motion abilities in complex mixtures. In addition, the insights in the influence of the interaction matrix on the molecular mobility also help to understand the contribution of "structural viscosity" to the viscosity of heavy oil.

show abstract

Section: Resultsmentioning

confidence: 99%

Diffusion‐Ordered Spectroscopy (DOSY) as a Powerful NMR Method to Investigate Molecular Mobilities and Intermolecular Interactions in Complex Oil Mixtures

Yuan,

Xu,

Yang

et al. 2023

ChemPhysChem

View full text Add to dashboard Cite

show abstract

“…These models can be trained using past data such as process conditions, feedstock compositions, and product specifications to understand complex relationships and provide precise predictions about the ultimate product quality [16]. Predictive models can be used at different stages of the oil processing, these steps include figuring out what kind of petroleum it is, making sure that the process settings are just right, and checking the quality of the finished product [17].…”

Section: Introductionmentioning

confidence: 99%

An attempt to promote intelligent petroleum processing: a look at how different programming languages works with deep learning models for gasoline quality detection

Alzahawi,

Issa

2024

Preprint

View full text Add to dashboard Cite

The current study explores how several programming languages affect deep learning models for gasoline quality detection, highlighting the significance of choosing the best appropriate language according to project needs. Featuring several scientific computations using an integrated development environment of the Visual Studio Code (VS Code) tool. This work has reached multiple outcomes: Python is known for its rapid development and adaptability, which are well-suited for exploring quality control techniques in petroleum products. Java is recognized for its stability and flexibility to work across many platforms, making it ideal for large-scale deployments that require dependable performance in refinery equipment. C/C++ is lauded for its speed and ability to manage hardware resources effectively, which is essential in real-time production settings, allowing for quick analysis and little delay in quality control processes. The study utilizes a deep learning architecture in Java, C, and Python, with frameworks such as Visual Studio Code to aid in model construction, with the goal of creating flexible tools for monitoring the dynamic quality of petroleum products.

show abstract

“…2023; 20 (12): 7036 2 of 17 Machine learning is a novel computational approach that is widely used in various fields of engineering. Appling this method to the available geomechanical data of an oil well can be a precise and useful manner for evaluating the effective parameters on the stability of an oil well wall in the petroleum industry [9,10]. But among of the regression methods, the XGBoost algorithm has strong advantages in finding the correlation between parameters and it can be optimized with many algorithms such Bayesian.…”

Section: Introductionmentioning

confidence: 99%

A Case Study for Evaluating Effective Geomechanical Parameters and the Influence of the Biot Coefficient on the Stability of a Wellbore Wall in the Asmari Formation using Machine Learning

Fahool,

Shirinabadi,

Moarefvand

2023

Trends Sci

View full text Add to dashboard Cite

In oil extraction projects, knowledge of reservoir geomechanics is essential for estimating the stability of wellbore walls drilled at great depths. In this regard, mechanics of porous media according to the definition of Biot’s coefficient instead of Terzaghi effective stress can provide a more accurate estimate compared to other analyses. Additionally, using artificial intelligence and machine learning algorithms such as XGBoost and optimizing it with algorithms like Bayesian, along with using SHAP algorithm as an interpretable AI model, can provide us with deeper insights into available data. In this research, 1st geomechanical data of a well in Asmari formation in southwest Iran was obtained through well logs and operational reports and then analyzed by machine learning. Also, a 1-way coupled reservoir rock-fluid model was built to investigate the volume of fractured rock around the well. The interpretation of machine learning results helps us better understand the parameters affecting instability in this well’s wall. Moreover, finite element model results indicate that assuming a value equal to 1 for Biot coefficient (or Terzaghi effective stress) leads to incorrect results and overestimates the volume of fractured rock around this well up to 13 times more than actual values. Therefore, any proper analysis regarding wellbore wall stability and evaluating effective stresses requires accurate knowledge about real and existing values of this coefficient due to simultaneous behavior of stress and pore pressure. HIGHLIGHTS One of the major challenges in the oil industry is the factors affecting the stability of the walls of oil wells during the extraction process, which identifying and addressing these factors can greatly assist in reducing extraction costs Nowadays, the collection of precise and extensive data through well-logging is a reliable source for assessing reservoir geomechanics, which, alongside machine learning can help us better understand reservoir geomechanics Given the conditions of hydrocarbon reservoirs, the mechanics of porous media, along with the definition of the Biot coefficient, can provide us with a better understanding of how shear failure occurs in well walls The XGBoost algorithm is useful for data analysis through regression and classification, as it can easily be optimized using the Bayesian algorithm and interpreted using algorithms such as SHAP Assuming a value equal to 1 for the Biot coefficient, in comparison to its true and in-situ value can cause many errors in estimating the stress-pore pressure interaction and the sanding rate of the well due to shear failure in it GRAPHICAL ABSTRACT

show abstract

Machine learning approach for predicting crude oil stability based on NMR spectroscopy

Cited by 11 publications

References 28 publications

Diffusion‐Ordered Spectroscopy (DOSY) as a Powerful NMR Method to Investigate Molecular Mobilities and Intermolecular Interactions in Complex Oil Mixtures

Diffusion‐Ordered Spectroscopy (DOSY) as a Powerful NMR Method to Investigate Molecular Mobilities and Intermolecular Interactions in Complex Oil Mixtures

An attempt to promote intelligent petroleum processing: a look at how different programming languages works with deep learning models for gasoline quality detection

A Case Study for Evaluating Effective Geomechanical Parameters and the Influence of the Biot Coefficient on the Stability of a Wellbore Wall in the Asmari Formation using Machine Learning

Contact Info

Product

Resources

About