Develop a Hazard Index Using Machine Learning Approach for the Hazard Identification of Chemical Logistic Warehouses
Loading...
Date
2019-11-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
With the rapid development of chemical process plants, the safe storage of hazardous chemicals become an essential topic. Several chemical warehouse incidents related to fire and explosion have been reported recently. Therefore, an accurate hazard identification method for the logistic warehouse is needed not only for the facility to develop a proper emergency response plan but also for the residents who live near the facility to have an effective hazard communication. Furthermore, the government can better allocate the resources for first responders to make fire protection strategies, and the stakeholders can lead to improved risk management.
The storage of hazardous chemicals in a warehouse is a complex problem. The potential hazards include flammability, reactivity, and interaction among different types of hazardous chemicals. Hazard index is a helpful tool to identify and quantify the hazard in a facility or a process unit. Various hazard indices are developed in history. However, the challenge for this research is to improve the current method with the novel technique to implement our purpose.
The first objective of this research is to develop a “Storage Hazard Factor” (SHF) to evaluate and rank the inherent hazards of chemicals stored in logistic warehouses. In the factor calculation, the inherent hazard of chemicals is determined by various parameters (e.g., the NFPA rating, the flammability limit, and the protective action criteria values, etc.) and validated by the comparison with other indices. The current criteria for flammable hazard ratings are based on flash point, which is proved to be insufficient. Two machine learning based methods will be used for the classification of liquid flammability considering aerosolization based on DIPPR 801 database. Subsequently, SHF and other warehouse safety penalty factors (e.g., the quantity of the chemicals, the distance to the nearest fire department, etc.) are utilized to identify the Logistic Warehouse Hazard Index (LWHI) of the facilities. In the last chapter, LWHI is applied to an actual case from Houston Chronicle, and several statistical analyses are used to prove that the LWHI is helpful for hazard identification to emergency responders and hazard communication to the public.
Description
Keywords
Hazard identification, Hazard index, Machine learning