Elsevier

Applied Soft Computing

Volume 37, December 2015, Pages 227-233
Applied Soft Computing

Estimation of surface tension of methyl esters biodiesels using computational intelligence technique

https://doi.org/10.1016/j.asoc.2015.08.028Get rights and content

Highlights

  • We developed MESTE for estimating surface tensions of methyl esters biodiesels.

  • Surface tensions of eight different classes of methyl esters were estimated.

  • Results of MESTE were compared with that of Parachor model and Goldhammer model.

  • Performance of MESTE outperforms that of Parachor model and Goldhammer model.

Abstract

Due to environmental benefits, methyl esters biodiesel got a considerable attention as a viable substitute to petroleum-based diesel. Surface tension plays significant role in atomization of this biodiesel since it controls the combustion process inside the engine through fuel–air mixing. Experimental determination of the surface tension of biodiesel is expensive and time consuming which limits its application as substitute for petroleum-based diesel. This is because proper choice of any methyl esters for diesel engine applications depend on the value of surface tension as high value of surface tension brings about difficulty in droplet formation. This work employs computational intelligence technique on the platform of sensitivity based linear learning method (SBLLM) to develop methyl esters surface tension estimator (MESTE) which estimates surface tension of methyl esters biodiesel with high degree of accuracy. Surface tensions of eight different classes of methyl esters were estimated at different temperatures by training and testing of neural network using SBLLM. The estimated surface tensions were compared with experimental results as well as surface tension obtained from Parachor model and Goldhammer model. The outstanding performance of the developed MESTE suggests its potential in estimating surface tension of methyl esters biodiesel for enhancing the atomization in biodiesels engine applications.

Introduction

As the demand for energy increases on daily basis, the need to compliment non-renewable energy sources remains crucial for meeting the future energy needs. Conventional petroleum-based fuels subject the environment to certain dangers as the greenhouse gases are released from these fuels. Biodiesels, in particular, methyl esters have received considerable attention as a result of its promising features such as renewability, biodegradability and environmental friendliness to mention but few [1]. With biodiesels of methyl esters, little or no modifications are needed to the standard diesel engines. Biodiesels may also be blended with petroleum-based diesel in varying proportion such as B20 (20% biodiesels, 80% petroleum-based diesel) and B5 (5% biodiesels, 95% petroleum-based diesel), depending on engines. The effects of some of these blends on standard diesel engines have been observed elsewhere [2] and the environmental impacts of their emissions have also been reported [3]. Methyl esters biodiesels are derived from animal fats and vegetable oils and consequently possess different compositions and physical properties such as surface tension that finally brings about different atomization characteristics which could affect combustion processes in engines [4]. Atomization is the first stage of fuel combustion in diesel engine in which fuel is converted to fine misted spray by allowing it to pass through a small opening under high pressure. Surface tension measures the force of attraction between molecules of fuel and biodiesel with high surface tension has molecules which are strongly attracted to one another and thereby making atomization and droplet formation difficult [5]. This creates incomplete combustion in an injection engine through disproportionate fuel–air mixing. Inadequate atomization of biodiesel reduces engine efficiency considerably. The proposed algorithm (SBLLM) allows accurate estimation of surface tension of any methyl ester so as to give insight into the appropriate one for a particular diesel engine application.

The force of attraction between the molecules of liquid (termed cohesive force) is mainly responsible for the surface tension. Therefore, molecular structure plays a significant role in liquid surface tension. The proposed algorithm uses molecular weight and temperature as descriptors in order to determine the effect of temperature on methyl ester biodiesel of choice since increase in temperature reduces the surface tension of biodiesels [6].

Experimental determination of biodiesel surface tension is time consuming and expensive which renders theoretical predictions and estimations essentials [6], [7]. Parachor approach is one of the frequently used methods of estimating surface tension of biodiesel fuel. It involves introduction of a weight factor (of each of the fatty acid methyl esters constituent of the fuel) to correct the error associated with Dalton type mass-average equation [9]. The added weight factor thereby complicates the method of calculation [8]. Another similar approach that estimates surface tension of biodiesel by adopting Dalton type mass-average equation emerged subsequently [4]. In this case, surface tension of constituent fatty acid methyl esters were obtained from the indices of topological structure which further requires weight factor in surface tension estimation. Another approach adopted in estimating surface tension is the correlation of excess Gibbs free energy with Wilson equation as well as surface tension to Gibbs free energy [8]. However, the obtained equation is quite complicated. Density gradient theory, perturbation theory and cubic equation of states are among other methods for estimating surface tension of biodiesels [6]. This present work develops a stable and less complex model that estimates surface tension of different kinds of methyl esters biodiesel at different temperatures using their molecular weight as input to the model.

This present paper introduces SBLLM in making excellent and accurate estimation of surface tension of several classes of methyl esters biodiesel at different temperatures using their corresponding molecular weight. SBLLM is among machine learning tools [10], [11], [12], [13], [14], [15], [16], [17], [18] that acquires pattern between the descriptor and target. Despite the fact that the model (SBLLM) is less complex, it saves computational time, achieves good stability within short period of time and eliminates local convergence. Among the uniqueness of SBLLM is its high speed in reaching the minimum error. This characteristic stands it out among other machine learning algorithms. It uses linear training algorithm for each of the layers of the two-layer feed-forward network [19]. The output of the first layer is assigned with random values and later updated using sensitivity formulas that utilize the weight associated with each of the layers. The computational time is saved while using SBLLM since its learnt weights through solving a linear system of equations which gives the corresponding local sensitivities of the least square errors with respect to input and output data points [19]. SBLLM has demonstrated its excellent predictive and generalization ability in tackling real life problem [20].

The results of our modeling and simulations indicate that the developed model is capable of estimating surface tension of any class of methyl esters with high degree of accuracy. The developed MESTE is characterized with high correlation coefficients (98% and 98.7% during training and testing phase respectively), low root mean square errors (0.558 and 0.496 during training and testing phase respectively) and low mean absolute errors (0.437 and 0.336 during training and testing phase respectively).

Section snippets

Description of the proposed method

SBLLM combines several unique characteristics which enhance its generalization and predictive ability. It considers the two-layer feed-forward neural network as made up of two different one-layer neural network in which Eq. (1) relates the input and output together [19].yjs=fji=0Iwjixiswhere j = 1, 2, 3 …, J, s = 1, 2, 3 …,S, I = number of inputs, J = number of outputs, wji = weight associated with neuron j, fj = non-linear activation number and S = number of data points.

The sum of the squared errors

Description of dataset

The dataset used in training and testing SBLLM algorithm consists of one hundred and five experimental values of surface tension of eight different classes of methyl esters biodiesel with their corresponding molecular weights and temperatures as drawn from the literature [3], [21], [22], [23], [24]. Statistical analysis was carried out on the dataset and the results of the analysis are presented in Table 1, Table 2.

The consistency and applicability of the adopted dataset are demonstrated from

Development of the MESTE using SBLLM

Development of MESTE involves training and testing SBLLLM algorithm with best parameters obtained through test-set cross validation technique. The accuracy of the developed MESTE during training and testing phase was recorded on the basis of correlation coefficient as 98.00% and 98.75% respectively. These high correlations are illustrated in Fig. 1, Fig. 2.

The developed MESTE also characterized with low RMSE and low MAE as presented in Table 4.

Comparison between estimated surface tension using MESTE and experimental values with the results of Parachor model and Goldhammer model

Fig. 3 presents the estimated surface tensions of

Conclusion

We developed methyl esters surface tension estimator (MESTE) that accurately estimates the surface tensions of eight different classes of methyl esters biodiesels using SBLLM with test-set-cross validation technique as the optimization strategy. MESTE is characterized with excellent predictive and generalization ability as demonstrated by the obtained correlation coefficient of 98.75%, low RMSE of 0.496 and MAE of 0.336 on the test set of data. The closeness of the results of MESTE with

Acknowledgments

The first author (Taoreed O. Owolabi) would like to appreciate Dr. Sunday O. Olatunji for taking him through the fundamentals of computational intelligence techniques.

The support under the project # R15-CW-11 (MIT-13104,13105) by KFUPM is highly appreciated.

References (27)

  • W. Yuan et al.

    Vapor pressure and normal boiling point predictions for pure methyl esters and biodiesel fuels

    Fuel

    (2005)
  • W. Yuan et al.

    Predicting the physical properties of biodiesel for combustion modeling

    Trans. ASAE

    (2003)
  • A.N. Shah et al.

    Effect of biodiesel on the performance and combustion parameters of a turbocharged compression ignition engine

    Pak. J. Sci. Ind. Res.

    (2009)
  • Cited by (16)

    • Artificial neural network, support vector machine, decision tree, random forest, and committee machine intelligent system help to improve performance prediction of low salinity water injection in carbonate oil reservoirs

      2022, Journal of Petroleum Science and Engineering
      Citation Excerpt :

      Nowadays, a large amount of both experimental and field data is being generated in diverse areas of petroleum industry, which has led to a widespread application of Machine Learning (ML) and Artificial Intelligence (AI) techniques for forecasting and optimizing field performance. A number of example applications are briefly presented here: Sensitivity based linear learning method (SBLLM) for prediction of surface tension of methyl ester biodiesels (Owolabi and Gondal, 2015); identification of lost circulation (Sheremetov et al., 2008); optimization of EOR processes such as waterflooding (Innocente et al., 2015); development of smart models using Multi-Gene Genetic Programming (MGGP) for evaluation of dew point pressure of gas condensate reservoirs as a function of reservoir temperature, composition, molecular weight (Kaydani et al., 2016); application of Artificial Neural Network (ANN) coupled with Imperialist Competitive Algorithm (ICA) to estimating oil flow rate (Ahmadi et al., 2013); application of type-2 Fuzzy Logic System (FLS) and Sensitivity Based Linear Learning Method (SBLLM) for permeability prediction in carbonate reservoirs (Olatunji et al., 2014), and application of ensemble model of Support Vector Machine (SVM) to quantify rock physical properties (Anifowose et al., 2015). Other recent applications include development of effective ANN, SVM and FLS models to forecast effect of acid fracturing on productivity of naturally fractured carbonate reservoirs (Hassan et al., 2021); ANN model development to help improve prediction capacity of propped hydraulic fractures to conduit fluids in gas shale zones (Desouky et al., 2021); genetic programming for development of generalized correlations to forecast thermophysical properties (i.e., viscosity, density, and solubility) of mixtures of solvent-bitumen with use in design of thermal-solvent bitumen recovery operations (Al-Gawfi et al., 2019); and smart models for prediction of performance of low salinity water injection in sandstone reservoirs using 1316 data points collected from coreflood data using black-box algorithms such as multilayer perceptron (MLP), SVM, and committee machine intelligent system (CMIS).

    • Molecular modeling of systems related to the biodiesel production using the PHSC equation of state

      2018, Fluid Phase Equilibria
      Citation Excerpt :

      Althouse [14], Bonhorst [15] and Rose [16,17] and co-workers were from the first scientists who stepped in this direction by determination of pure vapor pressure of fatty acid alkyl esters. After that, the investigation on the properties of pure fatty acid alkyl esters were continued by determination or calculation of vapor pressure, density, viscosity, speed of sound and surface tension [6,14,15,18–44]. Ortega et al. [45–53] studied VLE of several binary mixtures containing an alkanol and an alkyl (ethyl to butyl) alkanoate (methanoate to butanoate).

    • A hybrid intelligent scheme for estimating band gap of doped titanium dioxide semiconductor using crystal lattice distortion

      2017, Computational Materials Science
      Citation Excerpt :

      Hence, the need for non-linear modeling tool becomes necessary. The choice of the proposed SBLLM-GSA model is due to stability of SBLLM based model and its efficiency when available dataset is limited since experimental data are limited in this area of research due to difficulty in obtaining them [31,38]. Proper tuning of GSA parameters is of high significance for the purpose of circumventing premature convergence in GSA and subsequently improves local capturing and convergence rate so as to enhance the algorithm exploitation and exploration ability.

    • Modeling the thermal behavior of coal fly ash based polymer gel system for water reduction in oil and gas wells

      2017, Journal of Petroleum Science and Engineering
      Citation Excerpt :

      The unique features of SVR such as generalization of error bond and non-convergence to local minima widen its practical applications in several fields of study including petroleum industries (Akande et al., 2015a, 2015b). It has recently enjoyed wide range of applications in material sciences (Cekaite, 2009; Owolabi et al., 2015, 2016a, 2016b) and condensed matter physics (Owolabi et al., 2016c, 2016d) among others (Adewumi et al., 2016; Owolabi and Gondal, 2015). The choice of SVR as a modeling tool in this research work is due to the aforementioned uniqueness of the proposed model coupled with its reported stability as compared to other computational intelligence tools (Akande et al., 2015b, 2015c; Majid et al., 2010), (Owolabi and Gondal, 2017).

    • Computational intelligence method of determining the energy band gap of doped ZnO semiconductor

      2016, Materials and Design
      Citation Excerpt :

      It accurately estimates the desired target by aqcuiring a relation or parttern between the target and the descriptors (input to the model) [6]. Computational intelligence tools include support vector machines, fuzzy logic and evolutionary computation among others [7,8,9]. Among these tools, support vector regression (SVR) has enjoyed a wide range of applications due to its ability to model real life problems with high degree of stability and estimation generalization ability, even when developed using few data-points [10].

    • Application of computational intelligence technique for estimating superconducting transition temperature of YBCO superconductors

      2016, Applied Soft Computing Journal
      Citation Excerpt :

      Among the available optimization techniques in the literature include Cuckoo search algorithm [25,26], immune algorithm and Taguchi's method [27], Particle swarm based approach [28] and artificial bee colony-based approach [29] among others [30–34]. However, the optimization strategy employed in this present work for obtaining the best performance parameters associated with SVR was test-set-cross validation technique [35]. Detailed procedures for the test-set-cross-validation used in optimizing the SVR parameters go thus: for each run of generated training and testing set, values for the assessment criteria (correlation coefficient, root-mean square and mean absolute error) were monitored for a group of parameters C (bound on the Langrangian multiplier), λ (conditioning parameter for the quadratic programing (QP) methods), ɛ (epsilon) and kernel option (η).

    View all citing articles on Scopus
    View full text