Source localization for hazardous material release in an outdoor chemical plant via a combination of LSTM-RNN and CFD simulation

•A combination of Deep-learning and CFD is proposed for source localization of hazardous chemical leak.•The proposed model predicts the source location in a chemical plant that has complex terrain.•The proposed LSTM-RNN model predicts the true source location with 97.1% accuracy.•Even if the predict...

Full description

Saved in:
Bibliographic Details
Published inComputers & chemical engineering Vol. 125; pp. 476 - 489
Main Authors Kim, Hyunseung, Park, Myeongnam, Kim, Chang Won, Shin, Dongil
Format Journal Article
LanguageEnglish
Published Elsevier Ltd 09.06.2019
Subjects
Artificial intelligence
Hazardous chemical release
inverse problem
Neural network
Process safety
Source tracking
Hazardous chemical release
inverse problem
Source tracking
Process safety
Neural network
Artificial intelligence
Online AccessGet full text

Cover

More Information
Summary:•A combination of Deep-learning and CFD is proposed for source localization of hazardous chemical leak.•The proposed model predicts the source location in a chemical plant that has complex terrain.•The proposed LSTM-RNN model predicts the true source location with 97.1% accuracy.•Even if the prediction fails, a point of overlap within a radius of 10 m from the true source is derived. Chemical leak accidents not properly handled at the early stage can spread to major industrial disasters escalating through fire and explosion. Therefore, it is very important to develop a method that enables prompt and systematic response by identifying the location of leakage source quickly and accurately and informing on-site personnel of the probable location(s). In this study, a model that predicts the suspicious leak location(s) in real-time, using sensor data, is proposed. Feed-forward neural network and recurrent neural network with long short-term memory that learned the data gathered from the installed sensors are proposed to predict the Top-5 points in the order of highest likelihood. In order to train and verify the neural networks, the sensor data generated from computational fluid dynamics simulations for a real chemical plant are used. The model learns the inverse problem solving for accident scenarios and predicts the leak point with very high accuracy. [Display omitted]
ISSN:0098-1354
1873-4375
DOI:10.1016/j.compchemeng.2019.03.012