Process Plant Accidents Need New Tech: Predictive Analytics and AI

Process plants can be dangerous places, as we discussed in our first blog post detailing the primary safety concerns in process manufacturing plants. Despite the number of safety regulations in place, serious incidents still occur all too often, causing significant injury, major harm to the surrounding environment and wildlife, and often loss of life. 

Sometimes these incidents can cause ongoing damage which affects human, animal, and plants in the area for years. In the last 15 months alone, over 50 people have died, hundreds more have been injured, thousands of local residents were evacuated, and untold damage has been caused to plant and animal life due to explosions, fires, and leaks at refineries and processing plants. 

A Few Process Plant Safety Incident Examples

In May 2020, poisonous styrene monomer gas leaked from a chemical plant in Visakhapatnam, India, killing 12 people and injuring over 1,000 others. A year later, many still suffer from ongoing health issues due to exposure to the gas. 

Also in May 2020, 1 person was killed and 2 injured in a massive explosion at an automotive plastics factory near Naples, Italy, which demolished the entire factory. 

In July 2020, a chemical tanker exploded as it was being loaded with hydrogen peroxide at a chemical factory in Incheon City, South Korea, killing 1 worker and injuring 7 others. 

A blowout fire at a natural gas well in Assam, India, in June 2020 killed 2 firefighters, forced over 7,000 local residents to evacuate their homes, and caused immense damage to endangered wildlife at the nearby Maguri-Motapung Wetland and Dibru Saikhowa Biosphere Nature Reserve. Over a month later, 3 experts attempting to control the fire were injured. 

In December 2020, 8 people were injured in a fire at the Vindhya Organics chemicals factory in Hyderabad, India.

Four workers were killed in December 2020 at a wastewater treatment plant in Bristol, UK, when a silo holding biosolids exploded. 

Six people were killed and almost a dozen more were injured in January 2021, when a line carrying liquid nitrogen ruptured at a poultry processing plant in Georgia. 

At least 20 people were injured in a fire that followed an explosion at a chemical plant in Gujarat, India, in February 2021, and a similar but more deadly fire at a chemical plant in Pune, India, in June 2021 killed 18 people. 

In July 2021, a fire broke out at a chemical plant, this time in Bangkok, Thailand, after a tank containing styrene monomer exploded. 1 person was killed, 33 more were injured, and hundreds of nearby residents were evacuated for close to a week to avoid poison gas from the fire. Experts are concerned about ongoing air and water pollution. 

In the USA in July 2021, 10 people were injured in an explosion at a food processing plant in Kentucky, USA, when a liquid nitrogen leaked from a truck during unloading, and 2 employees died when acetic acid leaked at a chemical plant in Texas.

Also in July, 7 people died and 31 were injured at ChemPark chemicals plant in Leverkusen, Germany, when a solvent in a storage tank exploded.

Finally, in August 2021, employees who hadn’t received sufficient training broke safety rules while processing chemicals at a plant in Russia, causing a fire which killed 7 people. 

COVID-19 Raised the Incidence of Serious Safety Events

Experts noted an uptick in safety incidents as plants returned to operation following pandemic-induced lockdowns. It’s not clear what exactly caused the rise, but research shows that the danger of safety incidents is five times higher when shutting down and starting up a plant. 

That danger would increase further if the shutdown was unplanned and took place under rushed and pressured conditions. Sean Moran, an expert chemical engineer, commented: “I experienced a government-mandated plant shutdown myself in Panama after a covid outbreak there, and saw at first-hand how difficult it was to shut down safely and maintain safety under pandemic conditions.”

When plants restarted, they were also under a lot of financial pressure to return to a normal level of operations as soon as possible, which could lead to more shortcuts and greater risk-taking. 

The economic crunch due to the COVID-19 pandemic may also have induced some companies to cut their safety and maintenance budgets. Together with social distancing and work-from-home restrictions, most plants had fewer staff onsite to monitor equipment and processes if production was continuing in a limited fashion, or to track storage conditions of raw and partially-processed materials while the plant was closed. 

Unusual Times Bring New Need for AI

When AI, predictive analytics, and other cutting edge tools entered plants and refineries, their main role was to lower the cost of maintenance. But these terrible accidents highlight the need to apply AI solutions like predictive analytics to improve maintenance and safety. AI-powered predictive maintenance (PdM) solutions are particularly crucial when traditional labor-intensive, manual monitoring systems are disrupted. 

With fewer staff onsite and many processes and items of equipment shut down hastily, plants see the need to take advantage of AI, predictive analytics, and other cutting edge tools that can detect early signs of fouling in equipment that’s still in use, anomalies in ongoing processes, degradation of stored materials during a shutdown, and the buildup of potentially harmful gases. 

For example, the fatal gas leak in India mentioned above is thought to have been caused when the stock of styrene monomer self-polymerized. Because of pandemic regulations, there were no employees on site to circulate the material as needed to prevent polymerization, or to monitor temperature for signs that self-polymerization could be about to occur. Remote monitoring could have prevented loss of life and health for hundreds of people. 

What’s more, new stakeholders are entering the picture. Insurance companies have noted the potential for predictive analytics to enhance their risk assessment, since AI-powered predictions can be more accurate at predicting the likelihood of a claim being filed, and even how severe a potential claim would be. 

In a similar vein, equipment certification companies like TÜV SÜD, UL, Intertek and SGS are combining predictive analytics with their existing equipment monitoring and maintenance services. With the added input of AI, these companies can refine their ability to keep equipment up to date. At the same time, they aim to expand into the new field of validating the underlying AI algorithms that form the foundation of all AI-powered tools. 

Using AI to Improve Plant Safety

For every massive explosion at a plant or refinery, there are scores of minor safety incidents that lead to injuries and sickness onsite or pollute the surrounding environment, making AI safety solutions highly valuable. 

AI-based software like SAM GUARD aims to prevent a potential crisis by spotting the early signs of equipment failure or serious chemical buildup. SAM GUARD’s PdM as a Service solution, operated through a remote analytical monitoring service (AMS), can be up and running within days, offering the shortest route to consistent monitoring and anomaly detection from a distance. 

Remote monitoring with expert support means that companies can remain confident that they are doing everything possible to reduce the risk of safety incidents, even when operating with a skeleton staff. Advance alerts from AI solutions like SAM GUARD help companies to repair weaknesses while they are still minor, long before they could cause a chemical leakage or disastrous explosion. They also allow plants with social distancing restrictions to plan the best time to send in their maintenance team so as to avoid increasing risks of infection. 

Cutting edge predictive analytics and ML solutions also help by reducing the number of false alarms that plant supervisors have to deal with. Ensuring that the majority of alerts are valid helps stop workers from ignoring safety alerts that seem to permanently “cry wolf.” SAM GUARD minimizes the number of daily alerts to just the few that merit investigation, avoiding “alert fatigue” that can cause unnecessary harm.

Ongoing technology innovations such as robots are also succeeding in taking over many of the highly dangerous jobs in the field, reducing the risk of serious or fatal injuries to workers. For example, inspecting the inside of a fuel tank can be better handled by a robot. Other tech tools are working to minimize the environmental impact of manufacturing and process industries, reducing oil spills and leaks of hazardous substances that can affect plant, animal, and human health. 

Predictive Maintenance Can Improve Plant Safety

Every serious incident at a plant or refinery has a different cause, and frequently multiple causes. It can be very difficult to pinpoint a single change that would have prevented the accident, but that doesn’t stop R&D teams from working on new ways to use AI and other advanced technology for accident prevention. SAM GUARD’s sandbox features include root cause analysis to help with this challenging task of identifying the cause of plant incidents.

AI has come a long way from its use solely in equipment maintenance, and it still has a great deal of potential to actualize in the realms of safety, risk assessment, certification, and more. The COVID-19 pandemic highlighted the need for AI-powered predictive maintenance solutions to ensure that safety remains high even when normal operations are disrupted.

To learn more Request a Demo