This episode of the Pipeliners Podcast features a conversation with Dr. Faisal Khan from the Mary Kay O’Connor Process Safety Center at Texas A&M University. The discussion explores the intersection of artificial intelligence and process safety, examining how AI can enhance safety performance while addressing potential challenges. Dr. Khan also shares insights into his background, the mission of the MKOPSC, and the broader importance of treating safety as a science rather than just common sense.

Artificial Intelligence and Pipeline Safety Show Notes, Links, and Insider Terms

• Dr. Faisal Khan is the Department Head and Professor of Chemical Engineering at Texas A&M University, holding the Mike O’Connor Chair II and directing the MKO Process Safety Center and Ocean Energy Safety Institute, with research focused on offshore safety, risk engineering, and environmental system management. Connect with Faisal on LinkedIn.
  • Texas A&M University is a public, land-grant, research university in College Station, Texas, United States. It was founded in 1876 and became the flagship institution of the Texas A&M University System in 1948.
  • Mary Kay O’Connor Process Safety Center (MKOPSC) – A global center of excellence at Texas A&M University dedicated to advancing process safety through research, training, and industry collaboration.
• PHMSA (Pipeline and Hazardous Materials Safety Administration) ensures the safe transportation of energy and hazardous materials.
• AI (Artificial Intelligence) is intelligence demonstrated by machines in contrast to the natural intelligence displayed by humans.
• Generative AI is a type of artificial intelligence technology that can produce various types of content, including text, imagery, audio and synthetic data.
• PST (Pipeline Safety Trust) is a nonprofit public charity promoting pipeline safety through education and advocacy by increasing access to information, and by building partnerships with residents, safety advocates, government, and industry, that result in safer communities and a healthier environment.
  • The Bellingham Pipeline Incident (Olympic Pipeline explosion) occurred on June 10, 1999, when a gas pipeline ruptured near Whatcom Creek in Bellingham, Wash., causing deaths and injuries. Three deaths included 18-year-old Liam Wood and 10-year-olds Stephen Tsiorvas and Wade King.
• API (American Petroleum Institute) represents all segments of America’s natural gas and oil industry. API has developed more than 700 standards to enhance operational and environmental safety, efficiency, and sustainability.
  • Pipeline SMS (Pipeline Safety Management Systems) or PSMS is an industry-wide focus to improve pipeline safety, driving toward zero incidents.
• Process Safety – A discipline focused on preventing catastrophic incidents in industries involving hazardous materials, ensuring that engineered systems operate safely over time.
• Bhopal Disaster – A toxic gas leak at the Union Carbide India Limited pesticide plant in Bhopal, India (1984) that underscored the importance of process safety and led to advancements in chemical plant safety regulations.
• Pasadena Refinery Explosion – On October 23, 1989, a series of explosions caused by a release of flammable process gases used in plastic production at Phillips Petroleum Company’s Houston Chemical Complex in Pasadena, Texas. The incident resulted in 23 deaths and 314 injuries.
• Applied Research – Research aimed at solving current industry challenges, particularly in process safety, through laboratory and computational studies.
• Safety as a Science – The concept that safety should be systematically studied, developed, and implemented rather than relying solely on common sense.
• Training Programs – Educational initiatives at MKOPSC designed to disseminate process safety knowledge to industry professionals.
• Computational Safety Analysis – The use of advanced computing methods to model, predict, and improve safety outcomes in industrial processes.
• Fiduciary Responsibility – The ethical obligation of pipeline operators to ensure the safety of the communities they serve.
• Risk and Integrity Engineering – A discipline focused on assessing and mitigating risks in industrial operations to maintain safety and regulatory compliance.
• Regulatory Compliance – Adhering to safety laws and standards set by agencies such as PHMSA to prevent industrial accidents.
• Blind Spots in AI – Unintended consequences or risks associated with artificial intelligence that must be studied and mitigated in safety applications.

Artificial Intelligence and Pipeline Safety Full Episode Transcript

Russel Treat:

Welcome to the “Pipeliners Podcast,” episode 376, sponsored by EnerSys Corporation, providers of POEMS, the pipeline operations excellence management system, operations and compliance software for the pipeline operator to address safety program management, control room management, and field operations. Find out more about POEMS at enersyscorp.com.

[background music]

Voiceover: The Pipeliners Podcast, where professionals, Bubba Geeks, and industry insiders share their knowledge and experience about technology, projects, and pipeline operations. Now your host, Russell Treat.

Russel:

Thanks for listening to the podcast. I appreciate you taking the time. To show the appreciation, we give away a customized Yeti tumbler to one listener every episode. This week, our winner is Bryce Ruffier with Northwestern Energy. Congratulations, Bryce. Your Yeti’s on its way.

Learn how you can win this prize, stick around till the end of the episode. This week we’re going to speak with Dr. Faisal Khan from the Mary Kay O’Connor Process Safety Center at Texas A&M University, and we’re going to talk about artificial intelligence and process safety. Dr. Kahn, welcome to the Pipeliners Podcast.

Dr. Faisal Khan: Thank you very much for having me.

Russel: We’ve had the opportunity to have a few conversations and we’re leaning into some work together. I thought it would be great to have a professor and researcher come in and talk to us about process safety and AI. Before we do that, would you mind telling me a little bit about your background and how you came to be in your current position?

Dr. Khan:

Thank you first of all having me Russell on this. I can imagine all practitioners and be quite to keen, how an academic admissions to be on this podcast. Just to introduce the audience, my name is Faisal Khan and I’m currently head of the Department of Chemical Engineering at Artie McFerrin Department of Chemical Engineer at Texas A&M.

I’m also director of Mary Kay O’Connor Process Safety Center. That exactly is why I came all the way from the north to this beautiful, warm, nice, loving place. I born and brought up in India, and process safety is not an accidental choice for me. It is something that in fact, has caused me pain as I was growing up as a kid.

I born and brought up early in the northern part of India, not too far from Bhopal, where one of the major industrial accident took place. I was in grade four or five, I recall, as the incident took place.

As I grew up, I had the memory of how the newspapers and at that time, black-and-white TV channels have conveyed the tragedy that unfolded in the town and aftermath of that in the whole country.

[crosstalk]

Dr. Khan: …in my career…

Russel: What year was that, Bhopal?

Dr. Khan:

1984. A time came in my career when I have to do graduate study. It is that point I decided that perhaps I should be more meaningful in what I do. I decided in early 1990s to study process safety. This wasn’t a topic in any of the research institutions to be considered.

It was believed, and I am sorry to say, it’s still to be believed in a section of the society and community that safety is a common sense and not a science. People like me who took that common sense and said, “No, it has to become a common practice. To become a common practice, we do need to invest time resources in unleashing it to be a science.”

That’s where I spend my three decade of academic life, connecting the process safety to be a science rather than to be simply a common sense-based. I had my two decade of career up in the north, place called Newfoundland, Canada, where I was the Director of a very similar institute called Centre for Risk and Integrity Engineering. I was Canada Research Chair.

During COVID, as many changes have took place, this was a change for my life. From a land of frozen [laughs] scenarios, I was convinced to move to Texas, where the kids when I came in first, they said, “We directly moved from a deep freeze to an oven conditions.”

Russel: [laughs] You’ve gone from the frozen tundra to the hot sauna. That’s for sure.

Dr. Khan: Indeed. That’s a much better description [indecipherable 5:12] that could put it. That’s how I land up in Mary O’Connor Process Safety Center. That was my prime responsibility and that I still consider to be what else I do. That is my core expertise and the core interest. I’m glad to be here and happy to have this conversation.

Russel:

I have to thank Tristan Brown. Tristan has just left as the Deputy Administrator of the Pipeline Hazardous Safety Administration. He indirectly introduced me to you. He introduced me to somebody who I had a conversation with that says, “No. The right guy you need to talk to here is Dr. Khan.” I’m grateful to have had the opportunity to get to know you.

I thank you for sharing the story about Bhopal. I remember that quite vividly. That was a very, very significant safety incident. It impacted a lot of people. It was very, very severe. The other thing that you said I really want to echo is, you talked about not safety as common sense but safety as a science.

The other thing that I talk about is there’s safety that addresses slip, trips, and falls. That’s what people tend to think of more as common sense. Process safety is a safety science. It’s how do you build large, complex, engineered systems and do it in a way that they can be operated safely for the long haul.

You think about, the airline industry is probably a really good example of somebody who’s done a pretty good job of that. If you look at where the airlines were 40 years ago versus where they are now, incidents are exceedingly rare.

Of course, we just had a major one as we’re recording this, just last week in DC, a very major incident. They will get to the root of it and they’ll learn from that and they’ll go forward from there. If you would, tell me about the Mary Kay O’Connor Process Safety Center and what attracted you to move from Canada to the heat in Texas.

Dr. Khan:

I have to admit that as I was growing up as a researcher, in short, I will call it MKOPSC, which is stand for Mary Kay O’Connor Process Safety Center.

As a center of excellence, truly, in a process safety. This is a global center of excellence. When I was in early stage of my career, that is the center I look upon, seeing how they are making meaningful contributions, as I used to call, in the South, where I was up in the North, working on a parallel topic, more dominantly onto the pipeline and offshore industry.

I have known MKOPSC as long as they existed, as part of my earlier graduate studies, when they were in infancy stage and I was a graduate student, but as we grew up and all. Now, it is very interesting to understand and I believe, participants to know, those who are not aware of, that MKOPSC is an example of how some loved one can take a tragedy and create a greater opportunity.

Just like how I mentioned about Bhopal in 1984, unfortunately, 1989, a tragedy took place up here in Pasadena, in one of the refinery. Mary was a superintendent engineer. She lost her life along with her few friends and colleagues in that.

Mike, her husband, decided that he will not let her sacrifice her life to go away. He took that very painfully and cautiously in his mind and heart and decided that whatever settlement and resources he had, he will invest saving life. That’s the mission he took.

He reached out to many institutions. Perhaps he agreed upon to establish a center which is now known as Mary Kay O’Connor Process Safety Center, in her memory, with a single focus mission of saving life.

Mike was an integral part of the center. I had the blessing of working with him in early part here, in MKO. He passed away in 2021, in the late.

MKOPSC is truly an opportunity created through a tragedy which takes a mission to prevent such incident to occur. That’s how today’s MKOPSC, which is a reasonable-size applied research organization, is dedicated to establish and to achieve the goal of Mary and Mike’s like mission.

Russel:

It’s interesting to me. I’m going to bring up one other event that I talk about often, and that is the Bellingham incident. The Bellingham incident was a refined products pipeline in Washington state that ruptured, flowed into a creek bed and then caught fire, and killed some children. I know some people that were personally involved in that tragedy.

That community was awarded a settlement, and like the NKO, they took that settlement, and they used that money to establish the Pipeline Safety Trust. I’ll get a little emotional talking about this. I have a bit of passion about it.

What we do as pipeliners is important, and that we have a fiduciary responsibility of the communities we serve to make sure that we do that safely. It’s institutions like the Mary Kay O’Connor, Pipeline Safety Trust, and others that actually take and develop the research and develop the science, and it’s a very, very important work that is done there.

Dr. Khan: Absolutely, Russel. There’s not a single day when they talk about MKOPSC, and my eyes don’t get wet. It’s simply so deeply connected to all of us. We feel great pride in that mission that we are part of the venue where we could able to make an impactful contribution.

Russel: Absolutely. What are some of the things that the Mary Kay O’Connor Process Safety Center does? Can you talk about what capabilities it has, and what things it does to support industry?

Dr. Khan:

As you saw that how the inception of MKOPSC came in. We are a very closely integrated applied research center with industry. The governing body of the MKOPSC is a industry steering committee that actively guide, mentor, and provide direction to MKOPSC to ensure that our work remain most relevant to our mission, that is saving life.

How do we achieve that? In the three different facets, as I often call it. One is applied research, which generally are what we distinguish. Looking into the contemporary issues, what’s going on, and seeing how and where they can benefit us. We adopt these contemporary development and also provide guidance where we should be aware of. That’s a significant part.

A lot of our researchers, PhD students invest their time into it. Then the middle part is the training. We do share our knowledge, what we have acquired through our research in both lab and computational, to our industry partners in providing them these educational piece through their focus training, through their effective training program for the larger populations.

The third part, which is the services, where we could able to host different activities for a larger community in making them aware of the safety issues, safety challenges, and how to best address them.

These are the three facets we operate on, but working very closely with industry partners, understanding the ongoing challenges and try coming up with a potential solution so that they can study deeply and perhaps adopt these changes or solutions.

Russel:

The whole reason that I originally was seeking someone out to talk to with the Mary Kay O’Connor Center is the research activities and frankly, the capabilities and resources, not just through the Mary Kay O’Connor but also through the entirety of the Texas A&M University system.

There’s tremendous knowledge and tremendous resources there. Not just within the system, but through those relationships and connections elsewhere. Glad to have gotten connected and blessed to be learning more about what you guys do there.

I have roped you in to support a project, and I want to ask a couple of questions. In your opinion, what do you think the promise of artificial intelligence is as it relates to improving process safety performance?

Dr. Khan:

Russel, to be honest, this is a very promising technological development as we speak. This is revolutionary in many form and shape. There is a caution factor, like all other technological development, because they do come with their own associated blind spot which we need to better understand as we come. I’ll try to explain that in a historical fact to draw our attention to things.

When I was growing up as a kid, the computer revolution, the IT revolution come through, particularly in India where I born and brought up. India saw the greatest development of the IT sector, the whole software industries. At point, it was thought that everything will be replaced with the computers, and nobody will do anything, and all the computers will do.

Today, we are. Every household have the laptops. Every kid is all operationally efficient to run a laptop. In fact, our grandparents are able to run a laptop. We have learned the technology and adopt ourselves to it in a very meaningful way.

The fear factors, what we had up in 30 years back, all got resolved with up and downs and learning and we adopt. If you ask me, this is how I see artificial intelligence bringing in another revolution of a similar kind as PC in past.

Russel: I can remember conversations early in the days of personal computers where people would say, “There’s no way you need a computer in every house.” Now we have multiple computers for every person in the house.

Dr. Khan: Absolutely.

Russel: Need and provide value, not necessarily the same thing. I’d also say that the way we use and interact with computers is radically different than what we started with because I’ve been working with computers since there were no personal computers.

Dr. Khan: You come a very well-trained, Russel, but think about those in the offices. I’m talking about the corporate sectors or IT. You had a dedicated section who were running the computers. There were people who are specialized to run it. There is no such thing.

Russel: You had whole departments that were doing things that we all do ourselves now. [laughs]

Dr. Khan: Thank you. The software developer, the same way, people were writing the software and then now…

Russel: Nobody’s scratching out a memo and handing it to the typing pool, right?

Dr. Khan: That’s right.

Russel: We’re all doing it ourselves on a keyboard.

Dr. Khan:

We all could do. We all write our software. Think about that development which we all have seen, witness it, and being part of it. Artificial intelligence is one such development as we are all going to witness. We in fact have to start witnessing it. We are part of it, some intuitively and some not.

For a very classic example, I often say, we drive these days without [indecipherable 18:17] navigators. Initially, it was all the GPS and this and that. That’s all long gone. Now all we have, is our embedded in our smartphone, the GPS, and we rely them in like a small-g god in the unknown territory when we don’t know.

Russel: It’s true.

Dr. Khan: They usually guide us through very precisely.

Russel:

I’m involved in boy scouts. Try to hand a paper map and a compass to an 11-year-old boy and explain orienteering. When you did it when I was a kid, it’s a whole different thing than now. They know how to do it on their phone, but they don’t have the first clue how to do it on a piece of paper. Things happen. Things evolve.

You’re right. Artificial intelligence is a tool set. You made a comment though that’s really interesting. I want to explore that and talk about what you may be doing or what may be going on in your organization around this.

You said that it’s a new technology, and there’s going to be blind spots. We do not yet know what those are. How do you through research and such, discover those blind spots and mitigate any risk associated with that?

Dr. Khan:

Thank you, Russel. That’s the key point and the part of research I know. A large section of society is continuously working in adopting AI-based technology or AI into their operations, for all the good reasons, including some of the safety aspect, as we will talk about.

I believe you’re leading, which is a phenomenal effort in my view, one of the most revolutionary aspect of utilizing such brilliant tools to develop so that industry or pipeline industry especially, can able to learn and create scenario [indecipherable 20:12] .

Before we jump into it, let me give what are the blind spot in my limited mind and understanding. Because we are relying on an intelligent machines who help us in doing the task, and as the time will grow, they will become more independent.

They will become, what we call today, agentic AI. They will almost behave like me as an independent agent who is doing whatever task we expect it. If specialized task, which doing it.

Among other factors, which is widely discussed and in the literature and all regulatory framework, like ethics and other, the things which concern me from a safety perspective, what I call them, conflict between artificial intelligence agent and a human intelligence agent, which is a human being and an automated AI systems.

Like how we’re the human being, we have difference of opinions on the same issue based on our knowledge, based on our experiences, based on a perspective we all grow.

If we think about tomorrow that, let’s say, we have four agent and I’m the only supervisor. The agent are programmed algorithm, but as you see today, DeepSeek versus ChatGPT, they have a different algorithm to operate. For the same problem, they will view it differently.

Russel: That’s right. Like a human being.

Dr. Khan:

Exactly, like us. How do you converge these differences to a meaningful outcome. In human, we talk, we debate, we discuss, we push, we pull, and we find a common ground and move forward. That need to be considered and built into the machines. That is something which I believe is a very big blind spot as we speak today.

It might get developed over time, but from a scientific point of view, from an collaborative environment point of view, where AI will be working, that is one of the major black spot.

Russel: There’s another challenge with AI, and that is unintended bias where…because basically, an AI is limited by the algorithm you’re using, and it’s limited by the data that you feed to the algorithm and what you tell it about how to interact with that data, right?

Dr. Khan: Indeed.

Russel: Each AI is going to pick up the biases of its creator, like we do as human beings.

Dr. Khan: Absolutely. Thank you. All these biases are reasonable and expected, and they can drive individually if individual AI is the one to lead. The challenge we will see whenever we bring multiple AI agent or HI and AI, I mean. Artificial intelligence and human intelligence coming together to do something. That’s going to be the challenge.

Russel: I actually see a different challenge. More and more of the repetitive tasks are going to be done by the AI like more and more the repetitive tasks are now done by computers. The challenge will be that the human supervision is going to have a maturity of thought that is greater than the AI.

Dr. Khan: Thank you.

Russel:

As an engineer, one of the things you develop as an engineer by doing engineering for many, many years, is you develop a intuition about this is right, or this is wrong, or I haven’t looked at it hard enough, or I have looked at it hard. I know when I know, and I know when I don’t know, and I know where I’m somewhere in the middle, and I’m unsure.

I have a sense through intuition about that, even outside of doing the analytical work. We’re going to be asking younger engineers to be able to do that without the experience that others of us have gotten by doing it the old-fashioned slow way. I don’t know what that’s going to look like.

I’m sure there will be some great safety performance improvement possibilities in that. Also, there will be some new risks that are presented by that.

Dr. Khan:

Thank you, Russel. You hit the key point. That’s where I absolutely acknowledge that some of our repetitive human factor-related challenges, which often being seen as the causation of some of our safety challenges, will get reasonably minimized, or to a greater extent, minimized, because of the efficiency and accuracy we will be able to bring to the artificial intent.

Then, as I mentioned, it will become relatively rare, but that rarity will bring a consequences which will be far beyond our comprehension because when machine fails in that situation, it will be beyond our comprehension because we were all knowing what we were doing when we were doing it.

Russel: It’s likely to have a larger consequence. I haven’t thought about that, but that’s an excellent point. That’s absolutely right.

Dr. Khan: That is what make me sleepless when I talk about and think about it.

Russel:

If you think about human factors and operations, there’s certain things that we know about ourselves, about fatigue and confirmation bias and color blindness and other things that can impact how we process data and react to a situation. One of those things is when we do not see issues to address, we begin to think that there are no issues to address at all.

[laughter]

Russel:

If I go from having 100 alarms a day to having one a month, then do I know how to respond to an alarm? Then, how do I train for that? Anyway, this is all tee-up. I’ve done a few other podcasts with some other people talking about the PHMSA R&D project that we’re doing.

When I first had the idea for that and had written it up and submitted it to PHMSA, I met with you, and I sat down and I said, “Here’s what I want to do. Is this something you can help with?”

We were talking about combining gaming with generative AI in order to support doing training. One of the things that’s going to be important as we go forward is training is going to become more and more and more important.

Dr. Khan: Indeed. I would absolutely say, if we have to ask this question, “Where do we see easy and fast adoption of AI into the safety domain?” My two cent worth is training our people for the scenarios that they otherwise would not conceive, and that gives a way where we utilize.

Russel: Give them real-world experience of dealing with something that you might never see in the course of an entire career.

Dr. Khan:

Even we can’t conceive because of our…We as a scientist, my job I remember in very early stage, when I presented my study to my industry counterpart, they laughed at me and said, “What? Did you have the dream that this going to blow up here and that’s going to blow up there?”

Because this is all the scenarios as an engineer in this stage, and then you’re trying giving the numbers based on all scientific calculation.

As a practitioner, “What are you talking about?” they said, “This have been in operation last 15 years, and you’re saying this going to blow up just like that.” I understand their point, but they don’t understand that when the accident happen, this is indeed how it will happen.

Artificial intelligence-based capabilities give us this power to expand that, create the scenarios as what the project we’re talking about through the gaming technology, and provide our engineers to play with it and see what action of how frequency and level of engagement bring the mitigations or control of these scenarios. That is the real thing which they need to get trained.

I have to caution that by driving a simulated car in the video game doesn’t make you a good driver on the road. I absolutely agree, but that at least can make you aware of the potential hazard, make you feel of the steering and other disturbances, so when you come onto the road, you are well aware of.

It’s not, I take the word, but you’re not well trained, but you are well aware. A quick training will get you accumulatized.

Russel:

Oh, gosh. I wish I could find it, but I came across a YouTube video. There was a kid who had flight simulator, and he had been flying the flight simulator for the Cessna single-engine plane, and he wanted to fly a real plane.

A flight instructor sat with him, and he went through everything, and he was able to get the plane started, taxi the plane, take off, get into a flight pattern, and the instructor didn’t have to do anything until they were landing.

Dr. Khan: Beautiful.

Russel: The thing about landing an airplane is, if you don’t know what you’re doing, if you don’t have some experience with it, some repetition, then your nerves are going to come up because not all landings end well.

Dr. Khan: That’s right.

Russel:

One of the challenges is to try and get somebody as much of that experience as possible. I also think that using AI in a training domain has a couple of benefits. One, it’s low risk because we’re not actually injecting it into operations. Two, it allows us to learn about the AI and learn about how can we make this AI do two things.

One, give us a scenario that’s realistic, yet give us a scenario that’s rare and we’ve never seen before, and how do we build the data and train the model and set up the rules so that that’s the case? That’s a big part of learning.

Dr. Khan:

I would be saying the second point, to me, is a critical point, because where I sit and where I see working with my colleagues in a different industry, including heavily on the pipeline industry from the North in Canada and others, that these unseen scenarios are the hidden risk from a practitioner point of view.

From us as a researcher, we know like a medical doctor. They understand the risk of all potential misalignment, and that’s what they study, that’s what they’re trained for. As a living human being, we don’t unless we go through that situation, and the same true with us.

Artificial intelligence able to bring this rare scenarios into visuality that our pupil could be made aware and play with it to control and mitigate, and that is truly a remarkable feature and ability. In my humble view, if we built such capabilities to train our engineers, we for sure, enhance our operational safety significantly.

Russel: I absolutely concur. I’m very excited to have you working on this project with you. I’m glad we got introduced, and I’m excited to see what the future discloses.

Dr. Khan: Indeed.

Russel: [laughs] Look, thank you so much for your time. Really appreciate it.

Dr. Khan: No, pleasure is mine. Thank you having for me, and thank you for having me part of your wonderful program. I look forward working with your team and making a deeper impact through our collaborative effort. Once again, thank you for having me.

Russel:

I’m glad to have you, and you’re welcome. I’m sure you’re going to make a meaningful contribution for sure. I hope you enjoyed this week’s episode of the Pipeliners Podcast and our conversation with Faisal.

[background music]

Russel:

A reminder before you go, you should register to win our customized Pipeliners Podcast, Yeti tumbler. Simply visit pipelinepodcastnetwork.com/win and enter yourself in the drawing.

If you have ideas, questions, or topics you’d be interested in hearing about, please let me know on the Contact Us page at pipelinepodcastnetwork.com or reach out to me on LinkedIn. Thanks for listening. I’ll talk to you next week.

[music]