Hazmat Response and Artificial Intelligence: The Next Evolution

BY RICK EDINGER AND GREGORY G. NOLL

Employing risk-based response during a hazardous materials incident requires sound analysis from which to make good decisions.

• The COVID-19 Pandemic and Hazmat Response Programs
• Organizing the Hazmat Response: Five Models
• NFPA 470: The ‘New’ Old Hazardous Materials Response Standards

From the beginning of organized hazmat response in the 1970s, response teams have required good information to support the incident analysis process. Information sources can include on-scene observations (placarding, container behavior), the physical and chemical properties of the materials involved (safety data sheets), detection and monitoring results, and other factors that influence the behavior of a chemical creating an emergency. To be effective, this data collection and analysis process also requires synthesizing the information collected to create the incident action plan (IAP).

Hazmat Research Tools

Three decades ago, hazmat response teams often had a large cache of books and technical reference materials to assist in predicting chemical behaviors and synthesizing the collected data (photo 1).

 

As recently as the mid-1990s, it wasn’t unusual to see the hazmat planning officer (also called the science officer or the hazmat researcher) with three or four books spread out in front of him developing information and intelligence for the IAP. The “rule of threes” (consult at least three references) developed during this time, recognizing that there was no “one best reference source” and that technical reference sources sometimes conflicted with each other.

With the advent of laptop computers, much of the information these books and manuals contained was digitized, and hazmat response teams entered the computer age. Usually, the software programs were on disks, and maintaining databases required periodic maintenance.

More recently, higher-efficiency mobile applications on laptops and mobile phones have replaced the old, high-maintenance computer hardware and programs. What used to take up an entire shelf of space in hazmat vehicles you can now hold in the palm of your hand using various mobile apps. And, adding functionality, much of the detection and monitoring equipment manufactured today can communicate wirelessly with computer-based applications, thereby allowing for improved data collection and analysis.

This transition from the printed technical reference books to the floppy disks to mobile applications has made managing a hazmat incident safer and more predictable, providing the incident commander (IC) with more reliable, real-time data to support ongoing incident analysis, evaluation, and decision making. It is a common practice now to employ an Internet search engine during a hazmat incident, but acquiring hazard data and other information is only one part of the process. Validating the information and putting it to good use can be time consuming and challenging. Now, both society and the emergency response community are on the cusp of the next big transition that may be a game changer, artificial intelligence (AI).

Enter AI

AI—also called machine learning—takes the data and information collected from various sources and synthesizes it quickly to produce an outcome that would likely take a human being much more time and effort to produce. AI is being quickly integrated at higher levels into internet search engines and will soon be part of virtually everything that we do using digital technology. Millions of people already use voice-activated apps such as Alexa and Siri daily but may not realize that AI powers the voice recognition in these apps.

AI will not only automate many of the processes that humans currently manage but can do so with a foundation of data and information far beyond anything readily available to those without access to a supercomputer. Automating critical analysis to support more rapid decision making will be an early outcome of widespread use of AI. This technology is already writing resumes, passing standardized educational tests, and winning chess games and has submitted a successful college application to prestigious Stanford University.

So, how AI will assist with responding to and managing a hazmat incident? As an information-intensive endeavor, hazmat response could benefit greatly from the quick acquisition of synthesized information from multiple sources. We asked one of the new generative AI tools what AI will mean to hazmat responders. In 20 seconds, ChatGPT predicted that AI will be used for the following:

• Predictive modeling. AI can be used to create predictive models to forecast potential hazmat emergencies. For example, machine learning algorithms can analyze patterns in data from sensors and other sources to predict the probability and impact of a chemical spill or other hazardous event.
• Early warning systems. AI can be used to develop early warning systems that can alert emergency responders to potential hazmat emergencies. Using data from sensors, satellite imagery, and other sources, these systems can detect the presence of a hazmat release and notify emergency responders in real time.
• Decision support. Using AI can provide decision support to emergency responders during hazmat emergencies. AI algorithms can analyze data from sensors and other sources to provide real-time information to support the risk-based response process, such as the location, the container’s stress, breach and release, and the emergency’s severity.
• Robotics. AI-powered robots can be used to handle hazmats safely and effectively in emergency situations, thereby minimizing the risks of exposure to responders. For example, robots could locate and contain hazmats or remove contaminated materials from the emergency scene. Likewise, integrating detection and monitoring technologies with robotics and UAV capabilities could thereby increase responder safety.
• Simulation. AI can be used to create realistic simulations of hazmat emergencies to train emergency responders and improve their preparedness for real-world events.

Incident-specific questions that we have posed to AI technology have thus far only produced generalized responses. As with all emerging technologies, this will improve with time. And someone somewhere is likely already working on public safety-specific applications for this technology. Already, one company is developing AI-driven software for emergency management functions such as plume modeling, evacuation zone predictions, and shelter management.

Imagine the ability to use an AI application that can employ Dr. Ludwig Benner’s D.E.C.I.D.E. process, his GEBMO chemical behavioral model, and the Hildebrand & Noll Eight Step Process© while accessing thousands of chemical databases, integrating real-time incident data (weather conditions, chemicals involved, containers, and more), using geographic information systems specific to the incident location (topography, population), and getting IAP objectives—all within less than a minute. As AI matures and incident management applications develop, this vision is not too hard to imagine.

Transitioning from technical reference manuals to digital-based mobile applications has greatly improved hazmat response by providing ICs and hazmat officers with more reliable and real-time data to support decision making. The next big transition is the integration of AI into hazmat response. AI can be used for predictive modeling, early warning systems, decision support, robotics, and simulation. Its use in hazmat response is predicted to greatly improve response effectiveness and efficiency by quickly synthesizing information from various sources and providing ICs with a foundation of data far beyond anything readily available to those without access to a supercomputer. The future of hazmat response is exciting as AI technology advances and public safety specific applications are developed.

RICK EDINGER, EFO, retired as deputy chief from a large, all-hazards fire and EMS agency in central Virginia. He has served in public safety for 50 years as a firefighter, medic, and hazmat responder. Edinger remains active in hazmat response and is the chairman of the NFPA Technical Committee for Hazardous Materials Response Personnel. He has a bachelor’s degree in fire science technology from Columbia Southern University and is a graduate of the National Fire Academy’s Executive Fire Officer Program.

GREGORY NOLL, CSP, CEM, is the senior planning specialist for the South Central PA Regional Task Force and principal of GGN Technical Resources LLC, a consulting firm specializing in emergency planning, response, and incident management. He has 50 years of experience in the emergency response community and is the co-author of nine textbooks on hazardous materials emergency response and management topics. Noll is a member of the NFPA Technical Committee on Hazardous Materials/WMD Response Personnel and the Fire Engineering and FDIC International advisory boards. In 2019, he was inducted into the National Fire Heritage Center Hall of Legends, Leaders, Legacies and Legends.