Method for predicting viscosity properties of narrow fractions of lubricating oil base oils based on wide fraction composition

Abstract

The application discloses a method for predicting viscosity properties of narrow fraction of lubricating oil base oil based on wide fraction composition of the lubricating oil base oil, and comprises the following steps: collecting a base oil sample set, measuring hydrocarbon carbon number distribution data of samples in the sample set, and measuring viscosity properties of the samples in the sample set; establishing a viscosity property prediction model based on the hydrocarbon carbon number distribution data of the base oil; measuring hydrocarbon carbon number distribution data of a wide fraction base oil sample to be measured, drawing a carbon number distribution curve of different hydrocarbon wide fractions, fitting a hydrocarbon carbon number distribution curve of a narrow fraction base oil according to the hydrocarbon carbon number distribution curve of the wide fraction base oil, and obtaining fitting hydrocarbon carbon number distribution data of the narrow fraction base oil; and substituting the fitting hydrocarbon carbon number distribution data of the narrow fraction base oil into the viscosity property prediction model to obtain a viscosity property prediction value of the narrow fraction. The method realizes rapid measurement of properties of the lubricating oil base oil, and solves the problem of lagging research and development cycle caused by sample accumulation, cutting and property measurement.

Description

Technical Field

[0001] This invention relates to a method for predicting the viscosity properties of lubricating oil base oils, specifically, a method for predicting the viscosity properties of base oils with different boiling ranges and narrow fractions based on the broad fraction composition of lubricating oil base oils. Background Technology

[0002] Finished lubricating oils generally consist of 90% base oil and 10% additives. The composition and properties of the base oil directly affect the quality of the lubricating oil. With the increasing demand for optimal friction reduction in industry and machinery, the market demand for high-grade API Group III lubricating oils in my country continues to expand. Viscosity is one of the most important properties of base oils; it is a crucial indicator for classifying base oil grades and a direct basis for evaluating the viscosity-temperature properties of base oils. In lubricating oil production, different processing techniques often produce base oil products with a wide boiling point range (referred to here as "wide-fraction base oils"). To optimize the properties of the base oil, fractionation is used to adjust the ideal and unideal components relative to viscosity properties, thereby obtaining high-quality, high-viscosity base oils. In this process, the selection of the fractionation cut-off point is crucial. Traditional methods often involve cutting full-fraction base oils into narrow-fraction base oils, measuring viscosity properties with a viscometer, and then adjusting the fractionation cut-off temperature. This method has a long evaluation cycle, is not environmentally friendly, and makes it difficult to provide timely feedback of effective data to guide the adjustment of process parameters. Therefore, developing a method to predict the viscosity properties of narrow fractions of base oils based on the composition of the full-range base oil can better and faster guide the research and development and production process of lubricating oil base oils.

[0003] Currently, there are numerous reports on predictive models for important physical properties of base oils, such as viscosity and boiling range. Most of these models employ near-infrared spectroscopy and nuclear magnetic resonance spectroscopy, combined with chemometric methods. However, the characteristic spectral ranges of this type cannot reflect the relationship between the boiling range and viscosity of base oils, thus making it impossible to predict the viscosity properties of narrow fractions from wide-range samples using these models. With the continuous advancement of analytical techniques, the compositional analysis of base oils has penetrated to the molecular level. Establishing predictive models for the viscosity properties of base oils based on molecular composition data can effectively correlate these properties from a molecular perspective, potentially yielding more detailed and accurate information.

[0004] Chinese patent CN 110823764 A discloses a method for predicting the viscosity index of base oils. This technology first acquires the near-infrared spectrum of the base oil to be tested, establishes the correspondence between the near-infrared spectrum and the viscosity index, and calculates the viscosity of the base oil using a preset function, thus achieving an online analysis process that accurately predicts viscosity properties through near-infrared spectroscopy. This method and model are suitable for mature and stable process units, enabling online real-time analysis of the viscosity index of base oils. However, it cannot obtain the viscosity properties of narrow-range oils from wide-range base oils, making it unsuitable for research and development processes.

[0005] Xu Yupeng, Liu Dan, Chu Xiaoli, et al. Application of online near-infrared spectroscopy in lubricating oil hydroisomerization unit [J]. Acta Petrolei Sinica (Petroleum Processing), 2022, 38(3):729-738. This paper discloses the selection of representative base oil samples, collection of their near-infrared spectra, and the establishment of a near-infrared analysis model using partial least squares method to predict the distillation range, viscosity, viscosity index, and pour point of hydrocracking tail oil and base oil. The model has good accuracy and can achieve online analysis, possessing good practical and promotional value. However, this technology also cannot achieve the process of obtaining the viscosity properties of narrow-range oils from wide-range base oils.

[0006] Xie Xin. Simulation and Prediction Study on Antioxidant Properties of Hydrogenated Base Oils [J]. Petroleum Refining and Chemical Industry, 2020, 51(02):57-61. This paper discloses a 9-parameter neural network model for simulating and predicting oxidation stability by using multilayer perceptron neural networks and radial basis function neural networks, respectively, with the content of alkanes, cycloalkanes, alkylbenzenes, and viscosity index in base oil as input variables. During the model establishment process, the relationship between the oxidation stability of hydrogenated base oils and alkanes, aromatics, and bicyclic and tricyclic cycloalkanes was studied in depth, further clarifying the influence of hydrocarbon composition on the oxidation stability of base oils. This method closely integrates hydrocarbon composition data with viscosity properties, achieving the prediction of oxidation stability through model establishment; however, it cannot currently obtain information on the viscosity properties of base oils.

[0007] Wang Yanbin, Yuan Hongfu, Lu Wanzhen. Determination of viscosity index of lubricating oil base oil by near-infrared analysis method [J]. Lubricating Oil, 2001(06):53-56. This paper discloses a near-infrared spectroscopy-viscosity index correction model established using partial least squares correction. The model has broad coverage and can be applied to various crude oils, processes, and base oil grades, demonstrating good repeatability and accuracy. However, this method can only predict the viscosity index of base oil through near-infrared spectroscopy, providing relatively limited information, and it cannot obtain the properties of narrow-range base oils from wide-range base oils.

[0008] Chinese patent CN 112577987 A discloses a method for characterizing the molecular structure of lubricating oil base oil and the lubricating oil base oil itself. Starting from the molecular composition data of the base oil, it closely combines mass spectrometry carbon number distribution data, NMR structural parameters, and assumed rules of molecular structure to propose a variety of structural characterization parameters. This method can characterize the chain and cyclic structures of alkanes, cycloalkanes, and aromatics in the base oil, obtaining more structural characterization parameters of the base oil at the molecular level. However, this method focuses mainly on the structural characterization of the base oil and fails to effectively correlate molecular-level data with the viscosity properties of the base oil, thus failing to achieve a rapid prediction process of physicochemical properties.

[0009] Chinese patent CN 114446409 A discloses a method for calculating the viscosity index of lubricating oil base oil. It employs high-resolution mass spectrometry and nuclear magnetic resonance spectroscopy to analyze the base oil, obtaining information on the carbon atom distribution and average carbon atom distribution. Based on molecular structure information and the group contribution method, combined with the blending correction method, the viscosity at 40℃ and 100℃ is calculated. Finally, the viscosity index is obtained according to the established viscosity index calculation formula. This invention establishes a method for calculating the viscosity and viscosity index of base oil based on molecular composition data. The calculated viscosity and viscosity index are more accurate, demonstrating the close correlation between the carbon number information in the molecular composition data and the viscosity properties of the base oil. However, this invention focuses on calculating viscosity-related data from molecular composition data and cannot predict the viscosity properties of narrow-fraction base oils based on the molecular composition of wide-fraction base oils.

[0010] Viscosity is a crucial physical property parameter for evaluating base oil grades, and the production of high-viscosity base oils is a key focus of base oil research and development. Research and production processes typically involve obtaining wide-range base oil samples through processing. To obtain base oils with optimal viscosity properties, fractionation is usually employed to remove undesirable components. Therefore, this process involves fractionating the wide-range base oil samples and then measuring the viscosity properties of narrow-range base oils within different boiling point ranges to determine the optimal boiling point selection range for the base oil fraction. However, this process requires sample accumulation, fractionation, and property determination, with a single evaluation taking at least a week and consuming significant human and material resources. Therefore, it is necessary to provide a novel method for determining the viscosity properties of narrow-range base oils to overcome the shortcomings of existing technologies. Summary of the Invention

[0011] The purpose of this invention is to predict the viscosity properties of different narrow-fraction base oils after cutting by analyzing the broad-fraction composition of lubricating oil base oils. The aim is to understand the viscosity properties of base oil products with different boiling ranges without the time-consuming and laborious process of fraction cutting and viscosity property determination. At the same time, it can also realize the process of directly predicting the viscosity properties based on the base oil composition data, thereby guiding the production of high-viscosity and high-quality lubricating oil base oils.

[0012] To achieve the above objectives, the present invention provides a method for predicting the viscosity properties of narrow fractions of lubricating oil base oil based on its wide fraction composition, the method comprising:

[0013] (1) Collect base oil sample sets, determine the hydrocarbon carbon number distribution data of the samples in the sample set, and determine the viscosity properties of the samples in the sample set;

[0014] (2) Correlate the hydrocarbon carbon number distribution data of each sample with viscosity properties and establish a viscosity property prediction model based on the hydrocarbon carbon number distribution data of base oil;

[0015] (3) Measure the hydrocarbon carbon number distribution data of the wide-fraction base oil sample to be tested. Plot the carbon number distribution curves of different hydrocarbon wide fractions with carbon number as the abscissa and content as the ordinate. Fit the hydrocarbon carbon number distribution curve of the narrow-fraction base oil according to the hydrocarbon carbon number distribution curve of the wide-fraction base oil to obtain the fitted hydrocarbon carbon number distribution data of the narrow-fraction base oil.

[0016] (4) Substitute the fitted hydrocarbon carbon number distribution data of the narrow-segment base oil into the viscosity property prediction model described in step (2) to obtain the predicted value of the narrow-segment viscosity property.

[0017] The method for determining the viscosity properties of the sample described in step (1) is not particularly limited, and conventional methods in the field can be used.

[0018] Preferably, the carbon number distribution data includes data on the content of different hydrocarbon types and different carbon numbers in the base oil; the hydrocarbon types include alkanes, cycloalkanes with different ring numbers, and aromatics with different ring numbers; the carbon number distribution data is determined by mass spectrometry, and the mass spectrometer is preferably equipped with a soft ionization high-resolution mass spectrometer.

[0019] Preferably, the difference between the initial boiling point and the final boiling point of the wide-fraction base oil sample is greater than or equal to 50°C.

[0020] Preferably, the viscosity property prediction model based on the carbon number distribution data of base oil hydrocarbons in step (2) is established using the following modeling method:

[0021] Obtain carbon number distribution data for samples in the sample set;

[0022] The viscosity properties of the sample are measured, including viscosity at 40°C, viscosity at 100°C, and viscosity index. Preferably, the viscosity at 40°C and viscosity at 100°C are measured according to GB / T265, and the viscosity index is calculated according to GB / T1995.

[0023] The sample set is divided into a calibration set and a validation set;

[0024] Based on the hydrocarbon carbon number distribution data and viscosity property data of the calibration set samples, a calibration model for predicting viscosity properties is established using a multivariate calibration method.

[0025] Preferably, the boiling point range of the samples covers 100℃-700℃ and includes narrow-range base oil samples with different boiling ranges; the boiling point range of the calibration set samples covers the boiling point range of all predicted samples.

[0026] Preferably, the multivariate correction method includes multivariate linear regression, principal component regression analysis, and partial least squares method.

[0027] Preferably, the accuracy of the calibration model is verified using a validation set of samples.

[0028] Preferably, the fitted hydrocarbon carbon number distribution data of the narrow-fraction base oil in step (3) is obtained by the following method:

[0029] Using the boiling point corresponding to the carbon number of n-alkanes as the base oil fraction classification scale, the carbon number boundary conditions corresponding to different boiling points are determined;

[0030] Based on the hydrocarbon carbon number distribution data of wide-fraction base oil samples, carbon number distribution curves of different hydrocarbons were plotted with carbon number as the abscissa and content as the ordinate. The hydrocarbon carbon number distribution curves of narrow-fraction base oils were then fitted by Gaussian or Lorentz curves using carbon number boundary conditions to obtain the fitted hydrocarbon carbon number distribution data of narrow-fraction base oils.

[0031] Preferably, the n-alkane has a carbon number of C10 to C100.

[0032] Preferably, the viscosity properties described in step (4) include viscosity at 40°C, viscosity at 100°C, and viscosity index.

[0033] This invention, based on molecular composition data, establishes a model that can predict the viscosity properties of narrow-range base oils using hydrocarbon carbon number distribution data of wide-range base oils. This avoids the process of accumulating a certain amount of sample, cutting it, and then measuring the viscosity properties of the target fraction, which is often required in R&D and production. It enables rapid determination of lubricating oil base oil properties and solves the problem of delayed R&D cycles caused by sample accumulation, cutting, and property measurement. Furthermore, the method of this invention has smaller errors and more accurate prediction results.

[0034] The method of this invention, after obtaining wide-fraction base oils, directly measures their molecular composition data and then uses a model to predict the viscosity properties of base oils with different narrow-fractions. This method reduces the evaluation time to one hour and provides more detailed information, strongly supporting the research and development and production process of lubricating oil base oils. It solves problems such as delayed research cycles and high personnel and instrument costs caused by multiple tasks involving sample accumulation, cutting, and property determination. Furthermore, the method and model in this invention can be widely applied to meet the needs of different research institutions and production units. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the model establishment and property prediction process according to one embodiment of the present invention.

[0036] Figure 2 This is a comparison chart of the carbon number distribution curve of the narrow-fraction base oil obtained by fitting in Example 1 and the actual measured carbon number distribution curve.

[0037] Figure 3 This is a comparison chart of the carbon number distribution curve of the narrow-fraction base oil obtained by fitting in Example 2 and the actual measured carbon number distribution curve. Detailed Implementation

[0038] The present invention will be further described in detail below with reference to specific embodiments, but the present invention is not limited to the following embodiments. Any modifications that do not depart from the concept and scope of the present invention are within the scope of the present invention.

[0039] This invention collects base oil samples from different sources and with different boiling ranges. The molecular composition data of different hydrocarbon types and carbon numbers in the base oils were determined using a JMS100GCV gas chromatograph / soft ionization-time-of-flight mass spectrometer (GC / FI TOF MS) from Nippon Electronics Co., Ltd. The carbon number of hydrocarbon molecules in base oils is directly related to the boiling point and viscosity properties of the base oil, providing a more direct and accurate reflection of the viscosity properties. The mass spectrometer measurement conditions are as follows:

[0040] Gas chromatography conditions: The column was an empty tube column with a length of 30 m and a diameter of 0.25 μm; the injection port temperature was 350 °C; the injection volume was 1 μL, the split ratio was 10:1; the column flow rate was 2 mL / min; the temperature program was 60 °C held for 0 min, then increased to 300 °C at 20 °C / min and held for 5 min; the interface temperature with the mass spectrometer was 320 °C.

[0041] Ionization source and mass spectrometry conditions: filament current 40mA; emitter voltage -10000V; multichannel detector (MCP) voltage 2350V; ion source chamber temperature 80℃; mass scan range m / z 40~800.

[0042] Viscosities at 40℃ and 100℃ were measured using a Kaineng CAV42 fully automatic kinematic viscometer according to GB / T265, and the viscosity index was calculated according to GB / T1995.

[0043] Example 1

[0044] The viscosity properties of the narrow fractions at >400℃ and >420℃ are predicted by analyzing the hydrocarbon carbon number distribution of the wide-fraction base oil sample A.

[0045] First, a viscosity property prediction model based on the carbon number distribution data of base oil was established and validated.

[0046] The Kennard-Stone diversity method was used to divide the sample set covering 59 base oils into a calibration set and a validation set. Based on the carbon number distribution data and viscosity property data of the samples in the calibration set, a calibration model was established using partial least squares method. The modeling parameters and statistical results are shown in Tables 1 and 2.

[0047] Table 1 Modeling parameters and statistical results

[0048]

[0049] Table 2 Model Validation Results

[0050]

[0051]

[0052] The hydrocarbon carbon number distribution of wide-fraction base oil sample A was determined, and the results are shown in Table 3. Based on the wide-fraction carbon number distribution curve, narrow-fraction carbon number distribution curves for >400℃ and >420℃ were fitted using carbon number boundary conditions and Gaussian curve fitting. The fitting results are compared with the actual measured results after cutting. Figure 2 The fitted curves are similar to, but differ from, the actual carbon number curves after cutting. The main sources of error include measurement error of the carbon number distribution curve, fitting error, and error caused by the presence of other components during the determination of the full fraction curve, but the errors are within an acceptable range.

[0053] Table 3. Hydrocarbon carbon number distribution of wide-fraction base oil sample A

[0054]

[0055]

[0056]

[0057] The properties of narrow fractions were predicted using fitted narrow fraction carbon number distribution curves. The predicted results were then substituted into the calibration model, and the comparison between the predicted and actual results is shown in Table 4. The results indicate that the viscosity properties of narrow fractions predicted by the method of this invention based on wide fraction composition data are comparable to the actual measured values, and the error is within an acceptable range. This demonstrates that the prediction results of the method of this invention are good, and the property model based on molecular composition has a certain degree of fault tolerance.

[0058] Table 4. Predicted viscosity properties of wide-fraction base oil sample A corresponding to narrow-fraction base oil.

[0059]

[0060] Based on the high-resolution mass spectrometry analysis method and the established prediction model, the entire prediction process takes approximately 45 minutes.

[0061] Example 2

[0062] The viscosity properties of the narrow fractions at >280℃ and >460℃ were predicted by analyzing the hydrocarbon carbon number distribution of wide-fraction base oil sample B.

[0063] This embodiment uses the model established in Embodiment 1 for prediction.

[0064] The hydrocarbon carbon number distribution of wide-fraction base oil sample B was determined, and the results are shown in Table 5. Based on the wide-fraction carbon number distribution curve, narrow-fraction carbon number distribution curves for >280℃ and >360℃ were fitted using carbon number boundary conditions and Gaussian curve fitting. The fitting results are compared with the actual measured results after cutting. Figure 3 The fitted curves are similar to, but differ from, the actual carbon number curves after cutting. The main sources of error include measurement error of the carbon number distribution curve, fitting error, and error caused by the presence of other components during the determination of the full fraction curve, but the errors are within an acceptable range.

[0065] Table 5. Hydrocarbon carbon number distribution of wide-fraction base oil sample B

[0066]

[0067]

[0068] The properties of narrow fractions were predicted using fitted narrow fraction carbon number distribution curves. The predicted results were then substituted into the calibration model, and the comparison between the predicted and actual results is shown in Table 6. The results indicate that the viscosity properties of narrow fractions predicted by this method based on wide fraction composition data are comparable to the actual measured values, and the error is within an acceptable range. This suggests that the prediction results of this method are good, and the property model based on molecular composition has a certain degree of fault tolerance.

[0069] Table 6. Predicted viscosity properties of narrow-fraction base oil sample B for wide-fraction base oils.

[0070]

[0071]

[0072] Based on the high-resolution mass spectrometry analysis method and the established prediction model, the entire prediction process takes approximately 45 minutes.

[0073] In summary, the method of this invention, after obtaining wide-fraction base oils, directly measures their molecular composition data and then combines it with a model to predict the viscosity properties of base oils with different narrow-fractions. This method can shorten the evaluation time to one hour and obtain more detailed information, strongly supporting the research and development and production process of lubricating oil base oils. It solves the problems of delayed research and development cycles and high personnel and instrument costs caused by multiple tasks such as sample accumulation, cutting, and property determination. Furthermore, the method and model in this invention can be widely applied to meet the needs of different research institutes and production units.

[0074] Of course, the present invention may have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the claims of the present invention.

Claims

1. A method for predicting the viscosity properties of narrow fractions of lubricating oil base oil based on its wide fraction composition, characterized in that, include: (1) Collect base oil sample sets, determine the hydrocarbon carbon number distribution data of the samples in the sample set, and determine the viscosity properties of the samples in the sample set; (2) Correlate the hydrocarbon carbon number distribution data of each sample with viscosity properties and establish a viscosity property prediction model based on the hydrocarbon carbon number distribution data of base oil; (3) Using the boiling point corresponding to the carbon number of n-alkane as the base oil fraction division scale, determine the carbon number boundary conditions corresponding to different boiling points. Based on the hydrocarbon carbon number distribution data of wide-fraction base oil samples, plot the carbon number distribution curves of different hydrocarbons with carbon number as the abscissa and content as the ordinate. Combine the carbon number boundary conditions to fit the hydrocarbon carbon number distribution curve of narrow-fraction base oil with Gaussian curve or Lorentz curve to obtain the fitted hydrocarbon carbon number distribution data of narrow-fraction base oil. (4) Substitute the fitted hydrocarbon carbon number distribution data of the narrow fraction base oil into the viscosity property prediction model described in step (2) to obtain the predicted value of the narrow fraction viscosity property.

2. The method according to claim 1, characterized in that, The carbon number distribution data includes the content of different hydrocarbon types and different carbon numbers in the base oil; the hydrocarbon types include alkanes, cycloalkanes with different ring numbers, and aromatics with different ring numbers; the carbon number distribution data is determined by mass spectrometry.

3. The method according to claim 1, characterized in that, The difference between the initial boiling point and the final boiling point of the wide-fraction base oil sample is greater than or equal to 50°C.

4. The method according to claim 1, characterized in that, The viscosity property prediction model based on the carbon number distribution data of base oil hydrocarbons in step (2) is established using the following modeling method: Obtain carbon number distribution data for samples in the sample set; The viscosity properties of the sample are measured, including viscosity at 40°C, viscosity at 100°C, and viscosity index. The sample set is divided into a calibration set and a validation set; Based on the hydrocarbon carbon number distribution data and viscosity property data of the calibration set samples, a calibration model for predicting viscosity properties is established using a multivariate calibration method.

5. The method according to claim 4, characterized in that, The boiling point range of the samples covers 100℃-700℃ and includes narrow-range base oil samples with different boiling ranges; the boiling point range of the calibration set samples covers the boiling point range of all predicted samples.

6. The method according to claim 4, characterized in that, The multivariate correction methods include multivariate linear regression, principal component regression analysis, and partial least squares method.

7. The method according to claim 4, characterized in that, The accuracy of the calibration model was verified using validation set samples.

8. The method according to claim 1, characterized in that, The n-alkane has a carbon number of C10 to C100.

9. The method according to claim 1, characterized in that, The viscosity properties mentioned in step (4) include viscosity at 40°C, viscosity at 100°C, and viscosity index.

Images