Estimating the Dynamic Viscosity of Vegetable Oils Using Artificial Neural Networks (original) (raw)
Related papers
The surface tension is one of the main properties for characterization of the quality of the fuel atomization process for its use in a diesel engine. There is a lack of published information about the values of surface tension of vegetable oils. The aim of this research is to obtain a mathematical model based on physical properties that establishes a relationship between the surface tensions of different vegetable oils and their fatty acid composition. For this reason, from literature reports, experimental data of oils related to the surface tensions was collected. Knowing that surface tension as a function of temperature, a total amount of 15 oils from different feedstocks at 20°C was selected. The obtained models were developed based in the use of artificial neural networks and multiple linear regressions fits, based on the experimental data available in the literature. Also, the obtained models present a good correlation between surface tension and the fatty acid composition, with a 95 % of confidence interval and coefficient of correlation higher than 0,95. The coefficient of correlation obtained shown a high correlation between the analyzed variables. According to the obtained results, the proposed models are a useful tool for the surface tension estimation from the oils fatty acid composition.
Prediction of the Viscosity of Heavy Petroleum Fractions and Crude Oils by Neural Networks
Journal of The Japan Petroleum Institute, 1996
In this work, the prediction of heavy petroleum fractions was significantly improved by using a backpropagation neural network model. It was found that scaling the data, fed to the neural net, improved the convergence of the estimated parameter (viscosity) in reasonable time with acceptable accuracy. An absolute error of 3.4% was achieved which is found to be better than those by other conventional methods.
International Journal of Applied Information Systems, 2016
Artificial neural network as a modeling tool was utilized in predicting Viscosity index and Pour point of some bio lubricants from selected bio plants. In the laboratory the bio lubricant were extracted by Soxhlet processes and blended with additives from botanical plants after which their characterization were carried out. The neural model of Viscosity index and Pour point was developed based on parameters from Products extraction {Three (3) parameters}, Products blending {Four (4) parameters}, Products characterization {Two (2) parameters}. The nine (9) parameters were used as inputs into the network architecture of 9 (5) 1 2 in predicting the Viscosity index and Pour point, after series of network architectures were trained using different training algorithm such as Levenberg-Marquardt, Bayesian regulation, Resilient back propagation etc. Using MATLAB 7.9.0 (r20096), the prediction of the neural network exhibited reasonable correlation with the targeted (real) Viscosity index and Pour point and predicted Viscosity index and Pour point with the network errors being reasonable.
Artificial neural network modeling was employed to predict Viscosity index and specific heat capacity of grease lubricant produced from selected oil seeds. These oils were extracted from their seeds using solvent extraction method and characterized in Food Science Laboratory, University of Agriculture, Makurdi, Benue State of Nigeria. The neural model was developed to capture two groups of inputs data namely; materials formulation, and operating conditions. The effects of material formulation were represented by 5 parameters while the operation conditions were represented by 4 parameters. The neural network architecture BR 09 [5-4-3-2] 4 2 fitted the input/output relationship for the prediction of viscosity index and specific heat capacity; after series of training using different training algorithms. There were visual checking of predicted and experimental viscosity index and specific heat capacity which confirm that the artificial neural network model was successful in modeling the viscosity index and specific heat capacity.
Journal of Food Engineering, 2005
An artificial neural network (ANN) model is presented for the prediction of viscosity of fruit juice as a function of concentration and temperature. The fruit juices considered in the present study were orange, peach, pear, malus floribunda and black current. The viscosity data of juices (1.53-3300 mPa s) were obtained from the literature for a wide range of concentration (5-70°Brix) and temperature (30.7-71.7°C). Several configurations were evaluated while developing the optimal ANN model. The optimal ANN model consisted of two hidden layers with two neurons in each hidden layer. This model was able to predict viscosity with a mean absolute error of 3.78 mPa s. The performance of the ANN was checked using experimental data. Predicted viscosity using the ANN was proved to be a simple, convenient and accurate method. The model can be incorporated in the heat transfer calculations during fruit processing where concentration and temperature dependent viscosity values are required. This may also be useful in mass transfer calculations during filtration of the juice using membranes for clarification.
This study investigates the applicability of adaptive neuro-fuzzy inference system (ANFIS) and artificial neural network (ANN) approaches for modeling the fatty acid methyl esters (FAMEs) property including kinematic viscosity at various temperatures and the volume fractions of biodiesel. An experimental database of kinematic viscosity of pure biodiesel was used for developing of models, where the input variables in the network were the temperature, the number of carbon atoms (NC) and the number of hydrogen atoms (NH) of the composition of methyl esters (C8:0, C10:0, C12:0, C14:0, C16:0, C16:1, C18:0, C18:1, C18:2, C18:3, C20:0, C20:1, C22:0, C22:1, C24:0 were considered as input variables on the ANFIS and ANN. Moreover, the models are divided into saturated species from C8:0 to C24:0 and unsaturated species, from C16:1 to C22:1. The model results were compared with experimental ones for determining the accuracy of the ANFIS and ANN predictions. The developed model produced idealized results and was found to be useful for predicting the kinematic viscosity of biodiesel blends with a limited number of available data. Moreover, the results suggest that the ANFIS approach can be used successfully for predicting the kinematic viscosity of biodiesel blends at various volume fractions and temperature compared to ANN approach.
Prediction of palm oil properties using artificial neural network
IJCSNS, 2008
The palm oil industry in Malaysia has witnessed a prolific growth in recent years. For the past few decades, Malaysia has led the world in terms of production and export of palm oil. Therefore, physical properties and thermodynamic facts of palm oil have become one of the predominant parts in related chemical industries. Efforts to obtain physical properties of palm oil have been made in order to ensure the quality of the product. Experimental work requires time and is not economic wise. Predictions via correlation methods and thermodynamic models are not practical because the methods are less accurate and high cost as well. In this study, models of Artificial Neural Network (ANN) are constructed to study the physical properties of major and minor components of palm oil. The network is built in conjunction with the data obtained from literature and several journals. The network utilized a feed-forward structure with back-propagation algorithm. The physical properties estimated include the liquid density of palm oil, the vapor and liquid mass fraction of palm oil, and the vapor and liquid equilibrium of fatty acids. The major components consist of triglyceride and fatty acids while the minor components are carotenoid, tocopherols, and tocotrienols. Estimations of the physical properties of palm oil using ANN gives smaller errors compared to other alternatives. As an overall, the lowest Root-Mean-Square (RMS) error acquired for the physical properties of palm oil is less than 1% (0.01). These error decisions have pointed out the suitability of ANN for this study.
Prediction Of Nigerian Crude Oil Viscosity Using Artificial Neural Network
Petroleum & Coal, 2009
The viscosity parameter is a very important fluid property in reservoir engineering computations. It should be determined in the laboratory but most of the time; the data is not either reliable or unavailable. Hence, empirical correlations were derived to estimate them. However, the success of the correlations in prediction depends on the range of data at which they were originally developed in the region. In this study, artificial neural network (ANN) was used to address the inaccuracy of empirical correlations used for predicting crude oil viscosity. The new artificial neural network model was developed to predict the crude oil viscosity using 32 data sets collected from the Niger Delta Region of Nigeria. About 17 data sets were used to train the model, 10 sets were used to test the accuracy of the model, and remaining 5sets to validate the relationships established during the training process. The test results revealed that the back propagation neural network model (BPNN) were better than the empirical correlations in terms of average absolute relative error and correlation coefficient.
Chemometrics and Intelligent Laboratory Systems, 1996
Characterization of an unknown sample of edible oil and checking for its purity or for its possible blending with other edible oils is common problem facing the oil chemists. Commonly used methods in such cases are based on SFI, TLC and GLC. Earlier some techniques were developed utilizing the methodologies of statistical models like SIMCA. Methodologies based on artificial neural networks (ANN) were observed to be the ideal tools in many problems involving classification and prediction. A case study consisting of the GLC data of pure groundnut oil and 50%, 60%, 70%, 80% and 90% of its blends with five other commonly used edible oils viz., ricebran, mustard, sunflower, safflower and palmolein was undertaken. The training set data consists of the GLC peaks recorded at palmitic, stearic, oleic, linoleic and linolenic for the pure groundnut oil and 90% of its blend with other edible oils. The test data, on the other hand, consists of the GLC data of 50%, 60%, 70% and 80% of the blend of groundnut oil with other edible oils. The ANN model could successfully identify the constituent oils in the case of all binary blends of both training data set and test data set with 100% accuracy. Models based on multiple linear regression could successfully predict the exact percentages of the components of the composition of the constituent edible oils in the case of the binary blends.
A neural network model and an updated correlation for estimation of dead crude oil viscosity
Viscosity is one of the most important physical properties in reservoir simulation, formation evaluation, in designing surface facilities and in the calculation of original hydrocarbon in-place. Mostly, oil viscosity is measured in PVT laboratories only at reservoir temperature. Hence, it is of great importance to use an accurate correlation for prediction of oil viscosity at different operating conditions and various temperatures. Although, different correlations have been proposed for various regions, the applicability of the existing correlations for Iranian oil reservoirs is limited due to the nature of the Iranian crude oil. In this study, based on Iranian oil reservoir data, a new correlation for the estimation of dead oil viscosity was provided using non-linear multivariable regression and non-linear optimization methods simultaneously with the optimization of the other existing correlations. This new correlation uses API Gravity and temperature as an input parameter. In addition, a neural-network-based model for prediction of dead oil viscosity is presented. Detailed comparisons show that validity and accuracy of the new correlation and the neural-network model are in good agreement with large data set of Iranian oil reservoir when compared with other correlations.