Investigate & Digitize: A Forensic Approach to Modern Industrial Safety

“Almost everything I learned about process safety, I learned via an investigation.”

That’s the safety motto of Tony Downes, Director of Process Safety & Loss Prevention for Honeywell Forge and guest in Episode 11 of Forging Connections .  In this episode he shares industrial operational safety wisdom.

So, what does he mean by investigations?  When an incident occurs, Tony and his team review everything that went wrong. They go over every detail, from the environment of the incident to the failure point, to the data that preceded it.  He believes that the biggest safety mistake organizations make is to not take the time to run a proper post-mortem investigation.  Much can be learned – and prevented – when an incident is thoroughly diagnosed. (This goes for successes too! Investigate when safety projects work well, and you’ll be more likely to have repeated success in the future.)

Gathering Investigative Data

When it comes to investigations, data has much to teach us. Downes focuses on two strategies when dealing with data.

First, he recommends ensuring that the data is not only collected, but that it’s also easily accessible. If it’s hard to locate or analyze, there’s an increased chance that your safety staff will not be getting the full picture. Or worse, they may throw their hands up in frustration and abandon the investigation altogether.

Downes adds that a data barrier he has seen is when information is locked in systems that aren’t suited for the inclusion of analytics tools. (For instance, the computing system that controls the plant.) But there is a solution: store the data in a repository, often called a “data lake,” so it’s in a place that is more easily accessible and functional.

The second data strategy is making sure your team has access to the right data. This is a more administrative configuration issue where your need to make sure you're looking for the right information in the right place.

Downes, who has been in security investigations for more than 30 years, also says that when thinking about data, don’t just plan for how to use the data to investigate after something happens – data can also be used to help proactively predict when an incident might occur.

Watch for Warning Signs

He says for a large part of his career, the job was piecing together clues after an incident and he wanted to see if it was possible to look more proactively at the signs, because often you see hints that something is going wrong ahead of time. Today, Downes sees more of a proactive, predictive movement toward safety.

One of the driving forces behind this proactive movement is newer technology, like digitalization. Using smart technology can help focus an organization on predictive warning signs, instead of just reacting to things that have happened.

“Can we look for weaknesses in those systems? … It turns out that you can … You can find them ahead of time. You don't have to wait until the really bad thing happens.”

- Tony Downes, Director of Process Safety & Loss Prevention, Honeywell Forge

Environmental Safety

In Episode 11 of Forging Connections, Downes also talks about environmental safety and some simple but clever technical solutions he’s seen in that area.  One example is called Smart Flare – a solution for efficient flare monitoring and recording.

In oil and gas, a flare is the flame around a refinery. A requirement for environmental safety is to keep an eye on the flare, to ensure it isn’t producing smoke. If it smokes, it means there’s poor combustion, and in addition to the CO2, greenhouse gases are being emitted.

For decades, the solution used by refineries for watching a flare was to point a camera at it and have the operators in the control room check on it.  But operators have a lot of other responsibilities, and constantly checking the camera is tedious, unrealistic, and not efficient. What Smart Flare does is to insert an artificial intelligence (AI) box to watch the camera 24/7.

As its name implies, Smart Flare is smart enough to differentiate between a normal flame and when it's smoking, and it notifies the board operator when there’s an issue.  Plus, if there is an incident, the EPA will want to see the footage.  With Smart Flare, operators don’t have to go through hours or days of saved recordings to find the incident after the fact.  The Smart Flare AI technology flags any anomalies and automatically puts the recording in a special file for reporting purposes.

So, again, this is an example of a simple (but clever) use of technology that serves many purposes – it helps improve safety and operator efficiency while also assisting with environmental compliance.

Listen to the full podcast for further discussion about AI and the future of safety.)

Advice for New Hires

When asked what piece of advice Downes would give a new safety process engineer, someone just out of college and ready to start their first job, his answer had multiple parts of wisdom:

  • Seek out broad experiences.
  • Learn from the past, from other companies and incidents (the chemical safety board is a good source).
  • Network, go to conferences, and meet your peers at other companies.

While this advice is probably not so different than what other roles might hear, what is different is that safety engineers should focus on learning a lesson from each incident so you can motivate and educate your organization. Downes added, “the worst thing is not to learn the lesson from an incident.”

Listen to the Full Podcast

To hear an expanded discussion of these topics and more, listen to Episode 11 of Forging Connections for additional industrial safety wisdom.