http://lattes.cnpq.br/5872166081552556; NASCIMENTO, Claudemi Alves.
Résumé:
Hazardous extent predictions that ensure process safety while avoiding overprediction has been
an enormous challenge for hazardous area classification. Specific scenarios can be addressed
to build metamodels capable of predict rapidly and accurately hazardous extent. Therefore, this
work aims t o propose a model to predict h azardous e xtent in fla mmable a tmospheres b y
metamodeling construction and validation. Considering gas leakages in an open environment
and without obstructions, varying the factors storage pressure and temperature, orifice diameter,
molar mass, wind velocity and lower explosive limit. Different techniques were used to obtain
the metamodel, a s s imple l inear r egression, K riging mo delling, a nd a n interpolation mo del
based on w eighted harmonic mean. Every metamodel was built by u sing data obtained from
Computational F luid D ynamics (CFD) s imulation. T he s imple linear regression model w as
coupled with an axial decay velocity model using an algorithm to avoid the prediction errors
due to the nonmonotonic
wind effects. The suggested algorithm, called by modified empirical
model, calculated the hazardous extent value with a coefficient of determination (R2) equals to
0,9842 and a Root Mean Square Error (RMSE) of 0,0151 for a dataset of 600 cases of generic
gases release. The proposed model was also validated for 60 cases of hydrogen release, giving
a R2 equal to 0,9829 a nd a RMSE of 0,0680. A c ase o f s tudy c onsidering different r elease
scenarios was done, comparing the modified empirical model against analytical models and
with the IEC 60079101
(2020). The modified empirical model showed the best results when
compared with the other models, reflecting enough evidence to support the model accuracy in
predict hazardous extent.