CERAWeek: Digital Transformation: Creating value

Listen to Maana CEO Babur Ozden participate in a panel discussion with other industry leaders on digital transformation creating value at CERAWeek 2019.

 

Transcript:

Welcome everyone to our digital transformation creating value session here at CERAWeek, crashing into the end of the first day of CERAWeek, so I hope everyone’s had an enjoyable and engaging day.

We’re going to talk about a few things around digital transformation. It’s not as if technology is new to this industry and the energy space. I am reminded as a young fresh-faced Exxon geologist showing up my first day at work and being handed a box of colored pencils, and a few sheets of mylar and sent to a drafting table to go find oil and gas.

I think it’s safe to say things have come a long way since then, it’s quite a long time ago sadly, but this digital transformation; we really want to talk about and understand what that means. We’ll talk a little bit, I mean there may be some concern, it’s become the new big data, but I suspect there’s quite a bit more value in in that.

I couldn’t have a better panel that is assembled last second for this so we’re going to really look at this from really many different angles, and all valuable. So we have Saudi Aramco and Shell, that are going to provide their perspective with Ahmad and URI coming at it from you know a large EMP IOC sort of NOC standpoint.

Lal from Emerson is probably one of the leading service technical players in the automation space so your perspective on this will be fascinating, and then last and certainly not least, Babur Ozden who is the founder and CEO of Maana.

Maana you know really has done one of the pioneering software companies in this space. It’s good that a least a couple of our panelists have supported and invested in Maana so I think your perspective coming from that angle is quite valuable as well.

You know again, I think just to set the stage for this and why this is important now I think when we have been talking about digitalization lately we think about sort of the downturn and the fact that we’ve had to get more cost-efficient and I’m quite sure that a lot of the technology application and a lot of the effort we put into that was built around that need to get more cost efficient.

There’s something bigger going on here now too and that’s what we really want to talk about because   you know this energy ecosystem is nothing if not dynamic, and if anyone was in the session that was just here before it was really about the investment sort of environment around energy or the lack thereof and the challenges the energy industry is having and getting investment put into it.

I really think that this digital transformation is going to be a big part of putting the industry in the place it needs to be going forward, and I think the digital transformation concept is very much linked to the energy transition which we talk a lot about here at CERAWeek.

For the first time, there’s a bit of a horizon out there on things we took for granted maybe ten years ago in the exploration part of the business. Here’s one example; so what we do with the digital transformation and how it comes to be is quite important and what we really want to focus on gentlemen is the value aspect. We know with a lot of new technologies and new pushes there’s quite a bit of activity sometimes it’s a little bit fragmented.

But what we want to give back to this audience by the end of today is a sense of what value have we created and a little bit about maybe what we see looking forward. With that introduction, I mean let me just throw it to you, give us a little framework of where things are today with this digital transformation.

Dr: Ahmad O. Al-Khowaiter: I think the oil and gas industry has been a pioneer in digitalization from the 70s. We were the first to adopt digital control systems and that was really kind of like the foundation of what you might call digitalization today.

It is that interface, that physical interface, which we put a lot of investment in over the last 40 years. We invested in collecting the data and in storing the data over those years.

But we didn’t take the next step which was how do we use the data to make the best decisions possible, and I think that’s where we are right now. We’ve invested tremendous resources in the standards, in the equipment, and into collecting the data and actually dealing with the data and developing say, applications at the lower levels.

What is new today is where we bring that data to the top of the pyramid; where we make you know really strategic decisions based on the analysis of that data and that is the most difficult step. That’s where I think the tools that we were seeing that are coming out of other industries are most interesting.

You can say today we’re at the leading edge of data collection, data manipulation, perhaps, but we are not at the leading edge of databased decision making. This is where I think we are trying to find a way to leverage that huge data resource we have, whether subsurface or surface we have you know some of the largest data sets in the world.

I don’t think any other industry creates as much data as the oil and gas industry does today. I can tell you in our facility on our surface data alone, we have three billion data points per day generated by our facilities. Subsurface has similar numbers.

We have our field technology on the subsurface which collects all of our data and we utilize it. In I would say, isolated ways, how can we integrate all this data and make better decisions. That’s where I think we are at the point where experimenting with taking that data and using it for better decisions.

Examples are that we’ve done a lot of work in the past on processing seismic data and turning that into supporting say decision making but can we go to the next step? Can we take the cleanup that we’re doing with seismic data processing and turn it into interpretation?

There’s a difference between cleaning up and improving the data. Can we make the decision? That’s where the next step is. How can we do the interpretation now using automation using cognitive approaches? Similarly, on the surface, we have tremendous amounts of data around the health of our equipment; vibration data, all this great data and it doesn’t go beyond the pyramid.

It doesn’t go beyond the second or third layer of the pyramid, it just tells the operator in the field you’ve got a disaster out there, turn off the piece of equipment. That that doesn’t help me as a corporate leader to decide on what the next investment will be. So I think that data can help us make those decisions better.

I think there’s that next level of prediction That is the part that we are struggling with today, and we’re working with partners, in many cases you mentioned Maana for example. Many other companies, as well as large companies such as Emerson and others in trying to develop those first cases which will serve as the best examples that will then allow us to scale and adopt these techniques across our industry.

I think we’re at that really important turning point where we’re seeing the first results and that will give us the impetus to be able to scale. It’s that large amount of data that can be overwhelming in this problem because you quoted your three billion points and that’s a microcosm across the entire industry.

The trick of this is, is finding the opportunity that can drive a process or an application to a best-in-class using the right set of sensors. Then analytics have been ultimately enabling it to be scalable.

To be scalable, you have to drive a return. They have to have a value proposition associated with them, which is why what I believe in is starting smaller and scaling up is the real way to drive. This goes on a larger scale, on a Big Bang level.

It’s going to be hard to prove those returns and the dollars investments can be overwhelming. I mean this is already a theme there, and it’s a great segue for you Yuri. I think I mean there’s a lot that could be done. How do you look at the playing field and kind of decide where your priorities are?

Sebregts: Yeah, what we are trying to do is not have a very fixed plan over many years where we upfront decide exactly what we’re going to do and then methodically go step by step. We think that would be a great approach for a more mature technology or a technology that’s more incremental rather than transformational.

We do believe digital technology is one of those technologies that comes about that’s more transformational. Not the first time in the history of energy that a transformational technology comes towards us, so we have some historical practice.

But it’s not every other year and when it’s more deeply transformational, it’s very hard to see exactly how you would plan that out over a 10-year period. It’s more important to kind of step by step get stuck in try things out in the different places the way you operate and then pick the ones that are really successful and scale them up.

I do believe at Shell, where I’ve got the most detailed view of course, we are at that inflection point of having built very deep capability for the last few years 24, 36 months in terms of attracting people and building depths ourselves and getting key partnerships in place and trying things out.

Now we’re at the inflection point of firstly being able to really rapidly scale some of those first applications and roll them out across all our operations to give you an example if you take predictive monitoring an example of this surface data.

So I pointed out 12 months ago we would have been at building machine learning models for you know several dozens of pieces of  equipment based on a couple of thousand tacks at an asset and making that work and proving that it works and that the insights are better than any human at the refinery or platform can interpret the data.

Now we’re at a point that within a few months we can build that – more than 5,000 pieces of equipment and an entire refinery based on approaching a hundred thousand data points fully automated so then you’re at a point from trying it out yourself and being held back in pace by you know how fast humans can do it – really ramping up and scaling.

I think significant value is now coming soon.

Moderator: Interesting, so it does sound like we’re at the inflection point, you know sort of

where we’re early we’re in that stage, we’re starting to

see the value.

Karsanbhai: I think the returns David discusses are very immediate, if you find the right applications. As we’ve seen, whether it’s your heat exchangers or corrosion, whether it’s reliability, safety emissions, as we’ve talked about the returns on a small scale can be realized relatively quickly enabling them confidence in making a further investment as an organization.

I think that’s the beauty of digitalization. Once you find one payback on one case it’s very low cost to scale across. You know it’s basically the cost of the software, which is zero cost basically. It was an incremental sensor cost, exactly.

It really does scale well once you figure out what it is. The challenge with the scaling though is that it’s

very hard to generalize some of these models. The models of the approaches typically need to be trained or tested and in certain fists, and that’s where I think the biggest barrier is.

That’s what I think we need to train people. I think that’s probably a bigger challenge then the psychology itself, is developing our people to be digitally aware or digitally capable and then we’re asking people to do things differently.

Yeah work processes could be changed, we could have a lot of fun with that.

Moderator: So let’s segue down to bobber down there.

You know we’re talking about people, you’ve been going into these companies, providing a new technology, new software. How does it work? How does it roll into you know, a large corporation over the years? We have really come to learn the most important ingredients in digital use cases and each company’s size, Shell and Aramco is hundreds of them in their roadmaps this year, and following years.

It’s not the data, it’s the subject matter expertise that will guide the new models, new applications, to churn the data. The availability of those individuals to the development of their digital use cases are of paramount importance and it’s not just availability that they join a project once a week, but the tools and means they actually have.

They are the developers of these use cases so and I refer to this concept, I borrow somebody else’s statement of citizen data scientist which is an over exaggerated term, but ability to turn a subject matter expert whether it’s exploration or trading and anything in between is to engage with this digital platform and the data made available there to build these decision systems.

The large majors don’t have the shortage of the people nor the no half, it’s just that it’s the matter of time that when they do find time to actually tackle these roadmap of use cases.

There are certain ways and best practices we see these use cases are being developed, and they come to a point that they could hit that inflection rate and I was sharing in an earlier panel is that it would be after six years of seeing and being blessed in the middle of this transformation in very large oil and gas companies.

I think it is extremely important to make an assumption that two out of the three use cases will never see production time, so it’s extremely important to take that into account. That’s not a failure, but it’s the speed at which you find that.

As Ahmed was saying, that it will move the needle and then go crazy about it, so it’s that people are there. It’s just a matter of them making them available and making them own these use cases with easy to use products and tools.

I guess my perspective may be a question to Ahmed and Yuri. Every operation has challenging processes and as operators you know what they are. Whether it’s your heat exchangers, whether it’s your pipes, whether it’s your rotating equipment.

Those present themselves as the most viable immediate user cases, to try to use the analytic power of the data to improve, to drive up time, improve your safety, or reliability, or your equipment. I think the user cases in the industry are there and I think we understand them.

We struggle with the same things over and over again. I’m not so sure I agree with the challenge in identifying them. I think there’s low-hanging fruit, it’s readily available. But I don’t think it’s the identification of the use cases, so much as anybody in the industry.

After you know spending a couple of days thinking about it, comes up with more or less the same list of those cases. The skill to actually go and execute though, id a different thing, because it requires mastery of elements of technology but also requires mastery of treating data like an asset, which you know sounds easy but it isn’t always. That means you have to take care of the data. You have to maintain them, you have to make sure that the right people have access.

That the wrong people don’t have access, and if you’re truly going to embark on transformational business value, then you also have to change the way you work. I think especially, you know data analytics, essentially statistics on steroids, change is transformational what you do because it changes fundamentally what knowledge workers do.

Knowledge workers historically have spent the vast majority of their time doing diagnosis of the data, looking at the data, and using their own pattern recognition skill–which is called experience, and then figure out what the next step to go and do is.

We’re now at a point where we’ve proven that modern machine learning artificial intelligence is better than humans at diagnosing data and doing pattern recognition. That frees up a whole lot of time for people to no longer do diagnosis but and creative problem solution building and that switch has to manifests itself in your business processes otherwise you’re not really unlocking the value of these using isms and that domain understanding of how to do that takes a little bit more time.

Again, I think we are at the inflection point where we start to see that happening. How is it being measured even at this mid early stage? How are the returns being measured on a project-by-project basis? Is that a tough question? Then I’m going to follow it up like, is there such a thing as a digital transformation roadmap?

I mean there is a digital transformation roadmap for each company they have their own views, and I agree that it’s you know it’s more of an incremental type without a clear path.

We don’t really know where it’s going to lead to us but we know what it’s going to get us at the end of the day. I can take individual use cases. I’m not sure how much I can share of that particular Case, but I give you a case line familiar from many years ago when we started looking at prediction of certain measurements that are difficult to measure in the field. We looked at a challenge for us, which was the flaring challenge.

How can we minimize flaring across our whole system without replacing equipment? Basically, we wanted to cut costs. We looked at the data we had. We had massive data and we could infer from that, you know the actual flaring from the data.

That saved us a tremendous amount of money just being able to infer proper measurements from the existing data. That was a very early on use case which brought us down about 50% of our flaring. It was brought down just by that one simple application.

That’s a real number. We went from 100 approximate. Basically we were 0.5%, now flaring over sales, gas, or gas Production. We were at a little under 1% or a little over 1%. Sorry that’s it.

You know, if you think of the sale amount of gas that we produce per day, you know, approximately eight billion standard cubic feet a day of stay of gas that’s a tremendous amount of sales gas saved in value. A 0.5% is a tremendous amount of saving just from one application of the data. It’s basically an inference type algorithm that was used across to kind of infer the actual flared gas.

Once you report that, your operators change their behaviors and once they know what’s actually being lost and they start reducing. It’s those simple little applications which can take advantage of the data, put it available to the person who makes a decision, and you’ll see tremendous value savings from those kind.

There’s a lot of examples like that. I just used that one because I was personally involved in it many years ago and there’s many simple other decisions that are made that could be informed by good data analysis.

That’s where I think we have to explore. We have to change our cultures a bit. I remember another big challenge. I approached some of our experts in the seismic area and they really loved their rigorous models.

They don’t want to talk about you questioning or competing with their approach. They are seen as a threat in my view. You know I think that over time the subject matter experts have become to see this not as a threat but rather as a tool, and today you know there are there leaders who believe in cognitive approaches to the seismic interpretation.

I think the generation change is part of it. You know you have some older people who spent their whole careers and maybe developing their own models, but now today you have the younger generation which is more open to some of these approaches. Which I think is part of it. We need to get to the point where we are capturing the imagination of the younger generation and part of it is digitalization and AI approaches. They tend to get it.

I’ll give you one of my favorite examples here; the way we think about this obviously is around starting with a comparison of practice versus the best practice. The deltas that you have within your process or system, then the roadmaps and as you correctly said, everyone’s got a roadmap and every world map is prioritized differently.

The second steps are improving a value and the third, the fourth step, is the scalability. One of my examples of flaring exams of phenomenon is corrosion. Particularly when it comes to refining. It’s very easy to quantify; about 9 billion dollars spent by the refiners across the world in chemical inhibitors okay.

The way corrosion is managed and traditionally manages around using intrusive coupons and pipes, inset management processes, removing coupons, measuring the rate of corrosion, and a coupon, and then interpreting using a little bit of black magic in mathematics.

What if we had real time non-intrusive measurement of corrosion in pipes? What if with that amount of data we could actually generate a chemical inhibitor formula that was more accurate and could be used at a different frequency that would go at the nine billion dollars.

A 2% savings in your inhibitor cost is a real return to refine, so that’s the way we try to talk about visual transformation. The power of that data and translate into those analytics, that will drive an immediate return and can be scalable across many facilities.

So Shell, for instance you know projects- it’s smart people thinking and saying just well, you know this is a great application, let’s go try this! Maybe it goes through some review. Is there a process to how these new applications are being made?

It’s like with other technology development-it’s a funnel to be managed, so at this moment we’ve got more than 250 active projects that we’re working on. You know you work on them for a couple of weeks or months, and then you decide whether something has merit to continue or you kick it out into the long grass to decide and you focus on something else.

I see that continuing and some of those projects make it fairly fast to the end of the funnel and into deployment. That is then backed up by more underlaying larger milestone driven projects where you have to put infrastructure in place in order to be able to scale.

Some data systems we historically have in place where scaling is easy, but there’s other areas where we actually have to put infrastructure in place that your application sit on top if you want a two-track type of progress. It’s some longer-term planned large infrastructure projects and it’s lots of trial and error small projects on top of it.  Now every company here is going to be a little bit careful in quoting the exact numbers on how they assess the value. Value tracking is an attribution problem anyway in the end.

The technology enables us to do better business so it’s the businesses themselves that indicate your progress and how much of the better business results you didn’t attribute back to the technology. It’s a little bit you know, arbitrary.

What I can say, because of the ease of deployment. The low cost of investment. we are already making a lot more money from digital investment than what it costs so it’s self-funding value.

The value comes pretty quickly, and the value comes in doing things faster or you know a couple of percent better. The big value comes from the better decisions. There’s value in faster decisions. We can also now sort through seismic data like other people can using machine learning and we can find faults in seismic data for 20 bucks a compute time in half a day.

Previously it took geologists a couple of months to sort through the same, so you’re faster and therefore cheaper, but the big value is the better decisions.

We’ve got use cases where we’ve optimized the well location for a field based on artificial intelligence, and you quickly go into double-digit if not triple digit millions return on that particular technology investment.

That’s where the really big value is, yeah. In the core of the companies definitely to go for the long-term value, there’s a significant distinct activity to build digital technology stacks.

Which is sometimes fundamentally different from the way they tackle their traditional data storing, data processing, and doing traditional analytics. In this digital stack, the fundamental reason that the stack is built is to avoid the initial hardship of everything being done in silos.

Whether it’s your data or decision making. One of the added values I believe these initiatives are leaving these large organizations with is their digital ambitions. Not maybe in their traditional day-to-day world, but the digital ambitions, like a real digital company.

There’s hardly any side loss. Everything is in one place. Use cases are taking advantage of data brought from so many different places into one location. Sometimes it’s cloud, more and more it’s being selected as cloud, and then from there to run machine learning.

Machine learning, if you run it in your traditional world is an oxymoron. It’s a siloed world; you can’t run it end-to-end of your business you have no way to learn anything about your business through its data.

So, the requirement to run machine learning to be able to build its foundational algorithm. To build AI capabilities requires de-siloing of your data and your apps. Your use cases may be distinct, but they need to be sharing an intelligent core. The apps themselves are becoming intelligent as their output or their recommendations.

Apps themselves learning the efficacy of that relation. We see this type of it in the core applications where companies are going for big to build this kind of entire feedback loop into the system.

For that, you need this de-siloed environment for these things to run. That’s a very unique digital asset this company is a building that is new and not traditional or in their IT repertoire.

I mean I’m kind of amazed that it still always come back to data, but I guess what’s amazing to me a little bit is that the data is still a challenge. I guess there’s probably projects where it’s been pulled together and cleaned up in the right way, but I mean would you say that that is still an opportunity for the magical ultimate?

You know what I’m going say, buy more sensors. I think the problem of the silos is a problem we have to deal with as an industry. Especially, you know, there’s no question that what’s slowing down the value generation is these silos. Data security is a big problem.

I mean there’s a challenge, there are some real issues with data security that is preventing us from getting the most out of these applications.

 

Even within our organizations, there are data silos. I agree if we want to leverage data the way Google does, you need to see Google’s organization structure and how they deal with data.

It is very open, this is something you know that I think we have to learn from other industries. We have to understand what really needs to be secured and what can be shared and how we can change our learning process.

I agree if we don’t change the way we work, they’re just going to remain isolated use cases. We’re going to get minimal value from this, we have to change the way we work as companies and the way we learn as an organization.

It’s not just the application’s responsibility to learn, it’s really the people that are actually delivering the value at the end of the day. We need to change, I agree. We talked a lot about IT versus OTE and the barriers that traditionally have been there and their struggles in many cases organizationally to break those down.

There are also barriers to your point. For the few people out here that might not be operating, folks who may boast the information technology traditionally.

Folks known for technology skills well throughout their companies and have that tradition of doing that very well.

You know in many cases we’ve run operations in silos and we’ve incentivized people to focus on a certain set of production attributes that they need to meet efficiency attributes.

Reliability, uptime, etc and those worlds coming together has been a challenge. I think there’s two elements; there’s the cleaning up history, in terms of how do you deal with providing access to the data?

Historically, there wasn’t a big case for companies to have your trading organization access your retail customer data, or your B2B and B2C organizations to have the same data structure.

There’s this legacy of how that data is organized and that needs to be brought together. I think there’s a more fundamental thing that we are working through as an industry, because it’s transformational to how work gets done.

We’re also redefining how the players in the value chain work together and what everybody’s role is in that. Of course, some players are you know potentially more disintermediated than others by the new technology and therefore you start to get an element if everybody defending their data.

They can see there’s value in just owning the data and as long as they’re clear about how they’re going to play in the value chain in the future, they’re not open and won’t share it all and connect it all.

As an industry we have to work through those steps to see how the new models are working together. Then data structures and access follow that. Yeah that’s pretty amazing, how that kind of opens everything up then.

I mean it changes the relationships between groups, I mean the whole concept changes. As a supplier throwing out the levels of influence within these organizations to make a call on these programs, it can be very challenging.

Can we just touch on them? But just so you know they shall rename nameless, the big three? In in this new world of digital you know transformation we talked a lot about partnering here.

Obviously, there’s another aspect of the partnering. I’m just curious in a general sense, is that a positive thing, is it is it catalyzing, what we’re doing in this space? Has is it got any challenges?

I think partnering is really one of the best ways quickest ways to get to where we want to go with this. Much of what we are trying to do here, others have done in other industries.

So, the fastest way to do it is basically to collaborate with those who have implemented similar in other industries. We’ve partnered with you some of the bigger companies in this room.

We’ve also partnered with startups and venture capitalists like Maana and many others that we have seen a lot of success with. If you’re trying to do in-house on your own, you’ll probably end up with something that’s not really the most competitive solution, it’s better to open out and find out what is the market is going to bring and try to learn from that.

I’m a big proponent of venture capital, and also of piloting and collaborating with partners in trying to find these win wins for us. You can’t do it without partnering in my view. We cannot be competitive, it’s not our industry.

I have to say you know we’re not in the business of data manipulation. We are, but it’s only an end to a means. You know our goal at the end of the day is to produce our products at the lowest cost possible, and I think it’s best to collaborate and leverage everybody else’s.

I agree from an automation perspective we have application expertise, technology expertise but creating a ecosystem of partners, whether it’s analytics, various elements, that we can then bring united to the table to talk to our customers makes us more powerful.

We have a network of that, we’re creating and always looking to increase. I’ve got a slight nuance on that; we partner with all the three big tech companies to varying degrees and in different areas and with lots of startups and smaller companies as well more for the apps and with the big tech, for the kind of foundational layer.

I think what we’re working through is refining, what are those market standard solutions that we take from tech companies that are big in this and have more capability volume wise. Which are the niche things within that where we combine domain expertise with advanced digital skills to build our unique capabilities as companies so that we also have a have a role in all of that.

It’s finding that balance that makes you truly successful. I mean from your perspective again, you see, you know as you come in as a company is it different now than years ago, in terms of the acceptance of new technology and possibly partnering or working with others?

It is extremely different, and I think as a technology company, in a small company, I am very laser focused on where we could be value adding.

The difference is that the single biggest thing that stood in front of any major digitization campaign; it needs data to fuel it. To be able to take data from day to day transactional systems and make it day to day available to digital consumption has consumed rightfully maybe the first two three years of the digital infrastructure building.

That ability to make data available, to bootstrap your digital use cases, has been a

foundational investment layer. The moment the companies have passed that halfway through, or towards the end, it’s amazing how quickly they want to move.

My answer is yes, there’s a huge change, but change is driven from this investment into an infrastructure, to make their data available to their digital ambitions and digital projects.

All of a sudden, they’re not only ready to listen, but along that path they’ve created organizations, cross departmental teams to understand how to bring not only the data there, but what to do with it once you get there.

You’ve got roadmaps, some attempts to build long term roadmaps, and measures to look at what what happens after we do to three or four of those.

There is infrastructure ready. We see their people who are doing this in their day-to-day jobs.

They have their business unit leadership sponsoring and owning their people into these projects.

We see it among our client and potential clients. Industry has made its preparations, it’s now ready to go to the field and play, so they’re there. Fantastic, thanks everyone.

 

Talk to a MAANA Expert

Connect with Maana to learn how we can help you get answers to big questions.

Contact Sales

Strategic Partners

  • Accenture
  • Microsoft Azure

Learn More

Connect with Us

Stay in the know with the latest information about Maana services, events, news and best practices by email.