Flash point prediction of organic compounds using a group contribution and support vector machine

Lee, Chang Jun; Ko, Jae Wook; Lee, Gibaek

doi:10.1007/s11814-011-0164-8

Cited by 10 publications

(9 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To date, lower AAD values can only be obtained by considering models specific to restricted classes of compounds and involving boiling point or other experimental data. 25,28 An examination of individual fitted values reveals that the lower fit quality presently obtained for the training set stems from a small number of compounds associated with large errors. Let us have a look at the most severe overestimations, starting with the most significant one (+48 K) observed for hexadecyltrichlorosilane C 16 H 33 SiCl 3 .…”

Section: Resultsmentioning

confidence: 99%

Flash Points of Organosilicon Compounds: How Data for Alkanes Combined with Custom Additive Fragments Can Expedite the Development of Predictive Models

Mathieu

2012

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

An especially simple approach to the evaluation of flash point (FP) from additive fragment contributions is outlined. Being based on a square root expression derived from the examination of n-alkanes data, it avoids the need for nonlinear fitting procedures and for trial-and-error optimization of the analytical relationship between FP and molecular descriptors. Furthermore, in spite of a specially small number of additive contributions, the method can be applied to most organic molecules. For organosilicon compounds, it exhibits some advantages compared to previously available procedures, while providing very similar performances, with an average absolute deviation from experiment close to 12 K, a determination coefficient R 2 = 0.89 and only one error >40 K.

show abstract

Section: Resultsmentioning

confidence: 99%

Flash Points of Organosilicon Compounds: How Data for Alkanes Combined with Custom Additive Fragments Can Expedite the Development of Predictive Models

Mathieu

2012

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

show abstract

“…The GCM models predict the FP as a function of the number and type of functional groups which constitute a compound . In most accurate GCM models, artificial neural networks are exploited to map the relationship between the functional groups and the FP . Artificial neural network is one of the most efficient machine learning based tools for mapping the linear and nonlinear dependencies between variables and has been extensively used to predict various properties .…”

Section: Introductionmentioning

confidence: 99%

A Neural Network QSPR Model for Accurate Prediction of Flash Point of Pure Hydrocarbons

Mirshahvalad

Ghasemiasl

Raoufi

et al. 2018

Molecular Informatics

View full text Add to dashboard Cite

The present study introduces a QSPR model to predict the flash point of pure organic compounds from diverse chemical families. We used the Maximum-Relevance Minimum-Redundancy (MRMR) as an efficient descriptor selection algorithm to select 20 the most effective out of 1926 calculated descriptors. The selected descriptors and their combination with the normal boiling point data were used as model inputs and their correlation with FP was mapped using feedforward artificial neural networks. Study-ing various models, the best result was obtained by a neural network with 2 neurons in the hidden layer for which a combination of the selected descriptors and normal boiling point data were used as model inputs. Evaluating the performance of this model for a dataset of 727 compounds resulted in average absolute relative errors of of 1.36 %, 1.34 %, 1.44 % and 1.42 % and average absolute deviations of 4.48, 4.41, 4.75 and 4.66 K for the overall, training, validation, and test datasets, respectively.

show abstract

“…The closed cup method usually shows a few degrees lower compared to the open cup method . To predict the flash temperature of a pure substance, methods were applied based on QSPR and empirical simple correlation based on different properties of each material, such as boiling point and carbon number. − In the case of binary and ternary mixtures, because of the complexity of mixtures the flash temperature is predicted using methods based on activity coefficients ,,, or quantitative structure–property relationship and are calculated by modeling or methods such as Genetic Algorithm (GA) and Artificial Neural Network (ANN) or a combination of them. , Liaw et al in several articles predicted flash temperature of miscible and partially miscible binary and ternary mixtures by an equilibrium model. − ,− This model is used in many articles by various researchers. Moghaddam et al calculated flash point of ketones, organic acids, and alcohols by Liaw model and based on Uniquac activity model .…”

Section: Introductionmentioning

confidence: 99%

“…7−9 In the case of binary and ternary mixtures, because of the complexity of mixtures the flash temperature is predicted using methods based on activity coefficients 1,2,10,11 or quantitative structure−property relationship 12 and are calculated by modeling or methods such as Genetic Algorithm (GA) and Artificial Neural Network (ANN) or a combination of them. 13, 14 Liaw et al in several articles predicted flash temperature of miscible and partially miscible binary and ternary mixtures by an equilibrium model. 1−3,10−12 This model is used in many articles by various researchers.…”

Section: Introductionmentioning

confidence: 99%

Prediction of Flash Temperature of Binary Mixtures Containing Alcohol–Alcohol and Alcohol–Acid Using a Combination of Activity Models and Group Contribution Methods

Banihashemi

Movagharnejad

2018

J. Chem. Eng. Data

View full text Add to dashboard Cite

The present paper aims to examine the method used to calculate the flash temperature of binary mixtures containing alcohol−alcohol and alcohol−acid based on activity models of Wilson, UNIQUAC and NRTL. The binary interaction parameter is obtained by group contribution method in these equations. The advantage of this method is that the methods used to calculate the flash temperature are based on the activity coefficients and the binary interaction parameter are independent from VLE experiments and phase equilibria. These parameters may be obtained from group contribution methods for calculation of flash temperature of binary mixtures containing alcohol. Therefore, 12 binary mixtures containing 97 binary flash data were used; for the three models of Wilson, NRTL, and UNIQUAC, adjustable parameters were obtained by a similar method using group contribution method and it was later tested for three activity equations and 4 mixtures containing 33 binary flash data. Test results indicate that the overall average absolute relative deviation and average absolute deviation for normal butanol/normal pentanol, normal pentanol/propionic acid, and isobutanol/propionic acid when the fitted coefficients are used by equilibrium data are, respectively, equal to 3.6 K and 1.11%, 1.39 K and 0.43%, and 1.11 K and 0.34%. When the binary interaction parameters were obtained using group contribution method (the present work), the results were, respectively, 1.03 K and 0.33%, 1.02 K and 0.32%, and 1.11 K and 0.35%. Moreover for isobutanol/normal pentanol mixture with equilibrium data, the group contribution method was solely applied which was highly consistent with experimental data on flash temperature.

show abstract

Flash point prediction of organic compounds using a group contribution and support vector machine

Cited by 10 publications

References 22 publications

Flash Points of Organosilicon Compounds: How Data for Alkanes Combined with Custom Additive Fragments Can Expedite the Development of Predictive Models

Flash Points of Organosilicon Compounds: How Data for Alkanes Combined with Custom Additive Fragments Can Expedite the Development of Predictive Models

A Neural Network QSPR Model for Accurate Prediction of Flash Point of Pure Hydrocarbons

Prediction of Flash Temperature of Binary Mixtures Containing Alcohol–Alcohol and Alcohol–Acid Using a Combination of Activity Models and Group Contribution Methods

Contact Info

Product

Resources

About