Prediction of Flammability Hazard Properties of Chemical Mixtures using Computerized Tools

Fire and explosions are the dominant hazards in many industry sectors, especially oil
and gas. The management and control of such hazards is becoming increasingly
important from human, economic, environmental and legislative view points. One of
the important elements in the action is the hazard evaluation which involves the
prediction of the chemical’s properties which is flammability properties. Therefore,
the objective of this project is to predict the flammability hazards properties using
Quantitative Structure Activity Relationship (QSAR). This covers the prediction of
Upper Flammability Limit (UFL), Lower Flammability Limit (LFL), Limiting
Oxygen Concentration (LOC), Auto Ignition Temperature (AIT) and Flash Point
(FP).
QSAR models have been developed by using MATLAB software (version 7.8.0.347)
to estimate the flammability properties. The results from the developed models were
validated with other data obtained from experimental work and published literature
to confirm the accuracy of the developed models. Furthermore, the accuracies of the
developed models were compared with other QSAR models, which have been
developed by other researchers. The results showed that the accuracies of the
proposed models are the highest among other available models which are 99.63%
accuracy for UFL, 99.68% accuracy for LFL, 91.5% accuracy for LOC, 99.16%
accuracy for AIT and 98.42% accuracy for FP. A number of benefits can be
achieved if a computerized approach for hazard evaluation of chemical substances
could be demonstrated. Substantial savings in time and cost could be realized in
evaluating the hazard potential of new chemicals. Furthermore, it will give more
accurate data and all of the predicted values for each property will then be used in
the prepared spreadsheet in order to predict the properties of the mixtures.