(70a) Siri, How to Design a Procedure? Integration of Natural Language Processing in Procedure Design

Ade, N., Texas A&M University
Quddus, N., Texas A&M University
Kannan, P., Mary Kay O'Connor Process Safety Center
Peres, S. C., Texas A&M University
Mannan, M. S., Texas A&M University
Harhara, A., Texas A&M University
Parker, T., Texas A&M University
Procedures play a vital role in high-risk systems in ensuring that the required level of system performance is achieved while maintaining the associated risks below an acceptable level. Deviation from the procedures can lead to incidents with catastrophic consequences such as the Deepwater Horizon and Texas City refinery explosion. Effective procedure design can possibly reduce such deviations. However, currently the procedure design guidelines are subjective in nature and therefore, may be ineffective in helping the operators performing the procedures to achieve the required level of performance and safety. In the context of high-risk industrial environment, the relationship of the worker's performance with complex interactions among the workers, task, hardware and organizational aspects needs to be understood for designing an effective procedure as the procedure itself is an integral part of the industry’s operational architecture. This necessitates a sophisticated tool that can identify the complexities associated with the procedure execution and can also be scaled and applied effectively to thousands of procedures.

Natural language processing (NLP) is an artificial intelligence-based technique that explores how computers can be used to understand and manipulate natural language text for multiple applications. The technique has significantly evolved over the past decade as reflected in various commercialized applications such as Google Assistant, iPhone Siri, and Amazon Alexa. Interestingly, NLP has also been applied in safety domain in applications such as extracting constructive information from aviation safety reports and analyzing OSHA injury reports to identify patterns with respect to type of incidents.

In the present study, we are developing a novel NLP based algorithm that identifies the elements in a procedure that contribute to complexity. It will analyze the overall text and pinpoint the words and phrases that require the operator to make decisions and judgments at both step and procedure level. The algorithm will also identify whether sufficient level of information is provided in the step of the procedure for the operator to effectively perform the task. Lastly, the algorithm will recommend whether the steps in a procedure need to be split into multiple steps or combined into a single step to improve operator performance. An important aspect of the algorithm will be the ‘smart’ nature of the algorithm. The algorithm will continuously improve as it analyzes more procedures. The algorithm will identify complexity terms in a procedure, store it in its memory and look for same/similar terms in the next procedure. Thus, after applying the algorithm to multiple procedures, we will be able to identify text patterns from the memory of the algorithm that contribute to complexity in a procedure. This novel NLP based approach can be vital in deciphering the causes inherent in a procedure that lead to procedural deviations and eliminating these elements will help us to move towards a safer industry.


This paper has an Extended Abstract file available; you must purchase the conference proceedings to access it.


Do you already own this?



AIChE Members $150.00
Employees of CCPS Member Companies $150.00
AIChE Graduate Student Members Free
AIChE Undergraduate Student Members Free
Non-Members $225.00