Maurizio Pilu (00:04.294) 

Good day from London. I’m Maurizio Pilu and I’m the Managing Director of Safetytech Accelerator. I would like to welcome you to this episode of our Insight Series. Today we are going to talk about digital twins. Digital twins are something the industry has been talking about for a while and it’s gaining renewed interest on the back of this powerful wave of interest in AI that we are experiencing today. 

We are going to talk in particular about their use and benefits in heavy industries and critical infrastructure and how they can transform safety and risk management. Now, like other terms such as Internet of Things, Industrial Internet, Industry 4.0, and even digitization, these terms have over time started to cover an umbrella of things and even policies and government agendas, rather than something practical and solving problems of today. 

But it’s obviously far from that. So I’m quite keen in this episode that we perhaps try to demystify the term a little bit and bring it down to something practical that is relevant to people running operations. And perhaps discover that you might be using a digital twin already without even knowing it today, your plants or operational ships. So for this discussion, I would like to welcome our guest expert, Justin Anderson. 

 

Maurizio Pilu (01:29.894) 

Hello, Justin. And thank you. Hello. Thank you for joining us for this episode. So very briefly, Justin, can you please tell us about you and in particular what you’re doing right now in the context of digital twins? 

 

Justin (01:31.373) 

Hello, Maurizio. 

 

Justin (01:48.523) 

So I’m the director of the Digital Twin Hub, which is hosted at Connected Places Catapult. If I just jump back a couple of decades, in the late 90s I set up a software company and that company focused on building out systems around the Internet of Things. I then moved across to KPMG and became their global head of the Technology Center of Excellence at KPMG International before coming across the catapult. 

I’m here to do two things. One is to help guide industry to use Digital Twins as part of their operations and secondly to inform the national agenda as it pertains to policy and programs that sit around the world of Digital Twins. 

 

Maurizio Pilu (02:34.502) 

Thank you, Justin. Thank you. Fascinating. So let’s dive in, in the discussion that we’re going to have in the next 20, 30 minutes. So as I mentioned, I would like to try and unpack the term and bring it down to practicality. So let’s say you were in a pub, maybe not your average pub, maybe a pub where operations and safety directors meet after work. 

So you’re not your average pub. Suppose you were trying to give an elevator pitch about digital twins. And so how would you describe a digital twin in practical terms, in plain English? And its benefits to these people that are eventually, potentially the buyers and users of digital twins. 

 

Justin (03:19.333) 

Yeah. Okay, so in the pub, obviously I’m going to try and avoid any acronyms whatsoever or any complex technology and recognize that digital twins has become a recognized part of the technology world and in the same way as AI has. And of course, AI covers a multitude of different technologies as does digital twins. But I think it’s quite useful to see the two of them side by side. 

AI has primarily been inspired by that part of our brain, which is the firing of synapses between neurons and being able to create a level of intelligence from the inputs. Digital twins is a little bit like that part of our brain that connects to our sensory cortex, connects us to our eyes and our ears and our touch and our feelings, and allows us to get the information to the neural networks that can then process it. So AI by itself could be very smart. 

in its confined box. But in the real world, if you want good quality data, you need sensors to be able to pull that data in, make sure that it’s real time, and the digital twin essentially does that. So think of it as a part of the brain that allows us to connect to all those sensors, pull it in, and make sense of that information that we can then process. That’s my part. 

 

Maurizio Pilu (04:42.758) 

So let me try to even take it one step further in terms of practicality. So you mentioned sensors and you mentioned processing. Is it what it is really? I mean, it’s sensors on equipment and plants and then essentially a little brain, some AI that processes those sensors to provide insight. Is that all there is to it to some extent? 

 

Justin (05:08.391) 

Well, I’d like to reduce it down to something as simple as possible. I think when it comes down to it, what’s happening is the sensors are pulling information, typically analog information from the real world, converting it into something digital, bits and bytes, pulling that from wherever it is, being able to suck it into the brain. At that point, use that information to create a view of what’s happening in the real world. And that view, might be something you could visualize as a representation of the world around. I noticed that Google yesterday have just put out their latest augmented reality glasses. They’ll almost certainly be using digital twins to map the real world and pull them back into the view that they can create for a user that’s visualizing those. Once it gets there, we process information, and then the key that makes it the digital twin… 

is that that information is passed back to the real world, either to the sensor that might be an actuator as well, or a person responsible for being able to make a decision manually. But that flow of information from the real world to the brain and back again in order to drive some kind of change, that’s what a digital twin essentially is. Now, it’s a broad church and often people will talk about 

components within the digital twin world, different parts of it. And that’s usually how people start. They’ll build a little bit here, they’ll add something on, it will continue to evolve as more and more information is brought into it and a better picture is created. 

 

Maurizio Pilu (06:51.654) 

So I’ve been almost on a crusade in the past decade to try to unpack big labels such as AI and digital twin, internet of things. So if you were not calling what you describe digital twins, how would you call it? For example, is condition -based maintenance or predictive analytics a form of digital twins? 

Because what I noticed in our day -to -day work with clients and startups is that people are starting to call things a bit more for what they are doing rather than these big overarching terms. So for example, it’s predictive analytics of a plant or some piece of equipment, a form of digital twin, just to bring it out to terms that people really understand. 

 

Justin (07:42.563) 

You know, I think ultimately technology is there to help solve real -world problems and what you’re describing there is the fact that if you can understand before something goes wrong that it’s about to go wrong and take some sort of remediating action at that point, it’s going to save you a lot of money. So that’s the problem. Where does a digital twin fit into that? Well, let me give you a couple of examples. First of all, it may be…

that in fact you’ve got a device that’s attached directly onto the machine. It can pick up information about that machine, maybe vibrations, understanding how that machine normally operates. And when the vibrations change, it recognizes there’s a potential problem and then will create some kind of alert or intervention to say, look at me, I need some attention. And what’s actually happening there is that we have a sensor picking up vibrations. Those vibrations are then being interpreted into a model, 

which is essentially a representation of the normal running condition of that machine, which therefore is under the definition of a digital twin, performing that role of the virtual representation. It’s just not looking at it as a 3D thing. It’s looking at it as waves, sounds, vibrations, whatever it might be that is making that, that is picking it up. And then it makes the decision and then passes back some kind of instruction that could well be a predictive maintenance instruction. 

And so within that context, it’s a self -contained little device that can do everything. On the other side, you might look at London Underground and start to represent information across the whole tube network where the sensors will be pulled back into a large data center and they’ll be processed there separately. The twin of the underground will then be contemplating a broader variety of information from all sorts of different places and use that to determine what action should actually 

be taken at that point in time, which might be predictive maintenance, so it might be some way of improving the flow of the trains through the system. 

 

Maurizio Pilu (09:47.878) 

So you’re saying that perhaps digital twins are already pretty much in use in a lot of places. Maybe then just not called out. Would you say that using Google Maps is using a digital twin of traffic in a city? 

 

Justin (10:06.144) 

Yeah, well, that’s a great example. So if we think about that, the Google map knows your location, knows the speed that you’re traveling, combines that with speed that other people are traveling, takes that information to their central system that allows them to overlay the information they have about the networks that you’re traveling on, the roads, the rail, makes a decision about how to advise you about changing direction or avoiding traffic and passes that back to the car. 

Let’s say that you’ve got it integrated into the car itself rather than you navigating on your phone. And that allows you to then make the final decision to turn off at the next junction. But essentially, you sensed what’s happening, you provided that backup, you’ve got a virtual representation of the road, and the information is then passed back again. So yes, we all have digital twin technology sitting in our pockets. Many billions of people around the world are… 

already using digital twins every day to help them make decisions. 

 

Maurizio Pilu (11:09.03) 

Well, it’s arguably probably the most used digital twin on the planet right now. Well, it’s a very interesting point here. So behind the big labels, it’s a very practical solution that people are using every day in plants, in various operations to monitor bridges and so forth. It reminds me of a beautiful article on The Economist a couple of months ago. And it was a feature article, you know, cover page. 

It was entitled, Welcome to the Age of Boring AI. It was a compliment, actually, because the economist was arguing that behind all this flashing dystopian picture of AI and charge of PT used for doing homework, there’s actually a lot of pockets of very practical AI that industry and organizations started to adopt. And they’re not flashy. They don’t make the news, but they’re actually helping companies becoming more efficient. 

So are we perhaps entering a new phase here of the age of the boring digital twin, Justin? 

 

Justin (12:16.221) 

That’s a great question. I think that digital twins have been around for a long time. In fact, they’ve probably been around for as long as I’ve been around. I think, you know, arguably, if we look back to the Apollo program back in the late 60s, the first concept of a twin was in place there where we had this representation of the craft and decisions were able to be made by using that. And over the course of time, of course, what’s happened is we’ve got more processing power, more capability. 

to allow us to do more clever stuff. And we’re now at a point where that processing capability that we have in our machines has just reached a new level. Does that take it to a boring stage? Possibly, if it’s just there and we haven’t got to think about it. And we just utilize that processing power. And also importantly, that it works. Because of course, you know, when things work, we don’t have to think about them, do we? It’s when they don’t work, when you can’t do something that… 

 

Maurizio Pilu (13:07.43) 

it works. 

 

Justin (13:14.524) 

frustrating and you’re trying to address it but I think we’ve got to a stage with the software and the hardware, the processing power, the storage power that takes us somewhere that perhaps yes, now they’re just a bit boring. I like that. 

 

Maurizio Pilu (13:28.198) 

In a good sense, in a good sense. Just paraphrasing the economist, an excellent article that was. So much of our audience is interested in safety and risk management, but not only. But certainly they come from deep, heavy industry and critical infrastructure such as plants, energy, maritime, transport, manufacturing. 

 

Justin (13:34.843) 

Yeah. 

 

Maurizio Pilu (13:58.502) 

So can you picture for us for a moment, again, staying at a very practical level, Justin, with examples, if you can. So how digital twins are going to change how safety and risk is going to be managed in operations, OK? And what type is available right now? So let’s zoom in on safety and risk management. What is the benefit of modeling your assets and sensing and processing and? 

and having that kind of insight. Can you give us some examples? 

 

Justin (14:32.89) 

So you’d like me to dive into some sort of infrastructure and unpack it with a twin perhaps and see how this might work? 

 

Maurizio Pilu (14:41.03) 

Yeah, yeah, in particular around managing risks, you know, things blowing up and risk to people, risk to assets and the environment. So the risk management aspect, how can digital twins… 

 

Justin (14:51.29) 

Yeah. 

 

Well, so, I mean, what I would say in the first place is that as we move around as people, you know, we are dependent on our senses to ensure we don’t fall over, fall into something, burn ourselves on something. And we do that because we can see it, we can touch, we can feel it, we can smell. You know, those senses are essential to helping us manage the risk around us. Now, if you replace us with machines and systems, 

then having the senses about the real world to understand what might be about to happen and whether or not there’s some action that needs to be taken to avoid it is essentially the same thing. Now what does that look like in practice? Well, I’m currently doing quite a lot of work in the maritime sector, working with ports that are looking at their transformation journeys over the next few decades and trying to make sense of why digital twins would be important to that. And what I have seen is that there’s a very strong focus within that. 

world on safety. It becomes an absolutely core operating license for any port operator to be able to do what they do. So safety is critical. So a lot of the use cases that we’ve looked at will have a safety aspect to them to make sure that people are safe in the environment they’re operating in. And sometimes that will mean changing the work that people do. If there is high risk work and that can be automated or used. 

other sensors to identify information that needs to be used within the port operations, then the digital twin can play a role there, whether it’s reducing the number of pilots that are required to help a ship come into harbour, or whether it’s actually at the dockside, ensuring that people are safe away from the side of the harbour and are able to ensure that the operations reduce the risk for any individual. But of course, that’s at a personal level, understanding that risk. 

 

Justin (16:51.542) 

but risk goes much broader than that. We might be looking at the risk of flooding in the environment. We might be looking at how do we react at that point to ensure that our power remains on if the substation is under water. What happens at that stage? So we can look at many, many different types of risks that sit around an operation. And of course, every operator will have its risk register prioritised, understanding the impact of individual risks. 

What I’d say is that against the mitigations that are identified for any risk, looking at how adding in additional information from sensors that can then be processed in order and given the instructions and objectives to mitigate risk, I think it will become an increasingly important part of every risk management strategy across many different industries. 

 

Maurizio Pilu (17:51.878) 

Thank you for that. So you’ve taken a broad view of all the possible use cases. But if you had to pick, just to bring it down to practicalities, one, your favorite use case for which there is tech available right now, what would it be? What have you seen that people are already doing in ports in particular, for instance? 

 

Justin (18:14.164) 

Well, one of the projects, my favorite, I mean, I think, you know, it goes back to not being quite so boring, of course, because favorite is a bit more interesting in some respects. And if we see emerging technologies that are being used with new ways of being able to sense things, perhaps using drones or computer vision as a way of being able to capture that information, you know, I enjoy that because it allows us to go a little bit further than we’ve been able to go before. And we’re working on projects where… 

drones are inspecting assets and inspecting port sides and understanding what might be happening and a variety of different risks that might then be important to be able to prevent some kind of safety incident or some sort of security incident that might sit around that. So, you know, I’m focusing on the fact that that type of input is utilizing two important technologies there, the drones and computer vision. 

But to be behind the scenes, there’s some really boring stuff. You have to have the 4G, 5G networks operating across the port. You have to be able to have the right licenses for drone operators in the vicinity to ensure that you’re compliant with the various different authorities. So there’s all sorts of boring stuff that sits behind it. But the interesting, exciting stuff is how do you see further? What sort of power and sensing can we use that allows us to make… 

better decisions and simulate that environment in a way that allows us to make those decisions in an operational capacity faster. 

 

Maurizio Pilu (19:51.206) 

That’s excellent. I mean, computer vision and drones are also some of my favorites. So you’ve been telling us that digital twins are a bit of a Swiss knife to more efficiency, better risk management and so forth. But if they are so beneficial and so powerful, 

What’s holding back industry from adopting them even more widely? And I don’t mean industry in general, just because then it becomes a policy, you know, we say more collaboration, so yes, yes, yes to everything. I mean individual buyers. I mean individual buyers, they have to make a decision whether to, you know, sense up a particular equipment or monitoring an area or whatever. So is it return on investment that is not quite there yet or? 

 

Justin (20:21.554) 

What? 

 

Maurizio Pilu (20:47.206) 

clear, or is it the availability of the right tech? Does it sound too complicated? So what’s holding back people from making those individual decisions? 

 

Justin (20:55.984) 

I actually think that it’s unfolding at a rapid rate. It’s just unfolding in a way that is happening around us. You might not, in the same way as we might not recognize straight away, that Google Maps is a digital twin. Using GIS information across various different parts of our operations is absolutely key to the way that we’re pulling information. 

But I think that if you look at what we could, then the reason for the catapult, of course, is that we are here to try and break down the barriers and make markets work faster. And so, you know, we do look at some of the key barriers that if we could unblock them could accelerate even more the speed of change. But, you know, we get down to some fundamental challenges here. One is the skills. You know, people haven’t necessarily got the right… 

capabilities in some of the heavy industries. They’ve been very engineering focused and digital sort of sits on the side possibly and innovation team somewhere looking at some of the opportunities of turning some of that innovation into business as usual where you have quite different mindset between civil engineers and digital engineers. Digital engineers operate in an agile world with iterative updates to systems and a continual evolving process. 

they don’t have to have the entire system designed in the same way as you do if you’re about to build a ship and you haven’t thought about all the different elements of the ship before you start. The digital world is able to work in a different time scale and is able to make changes along the way and learn from the real world that it operates in. And so that iterative, agile, continual, progressive change is what we’re seeing. As opposed to the… 

perhaps the large -scale vendor of a digital twin solution that is selling a high -end enterprise digital twin as the answer. I personally don’t think that is the way we’re going to see things unfolding. I think we will see multiple services, digital services, operating bit by bit. Each of these components start to come together. But every component itself is likely to have to go through some kind of… 

 

Justin (23:19.149) 

individual cost benefit analysis. What if we add this on? Where’s the value going to come back? So we don’t have to look at the value of the whole system. We look at the constituent parts. We understand what happens if we add something to that. And when we do that, we’re not just looking at the individual use case, another use case, and the value of that use case. We’re looking at incremental value that comes from adding that use case to all the other use cases. What does that really mean? 

it’s a bit like a fax machine. A fax machine has only got so much value if there’s only one of them in the world. As soon as you’ve got a network of fax machines talking to each other, they become actually quite a valuable communications channel. In the world of digital twins, the more digital twins you can connect, the more data flows around the systems, or the system of systems, the more value will be created as well. But it will be very incremental. It’s a journey. 

 

Maurizio Pilu (24:14.374) 

Yeah, and this, yeah, I will tend to agree with this view. And it goes back to that economist article of the Boring AI. And I think the article was advocating that very few companies are going to become the AI enterprise overnight with a mega project. It’s going to be more like the incremental aggregate of a number of applications that are becoming smarter and smarter, that are a bit more manageable. 

You can create a use case in isolation. You can really control the implementation and so forth. That said though, Justin, it’s one of those things that I, you know, in my introduction talking about these big labels that tend to grow and grow and grow and cover everything. I remember when Internet of Things was hot, still is, and actually changed name. Now it’s become real. It’s just become boring now. 

Sensors are everywhere. But when it was hot as a label, I remember having a heated argument with somebody arguing that a PC connected to an ethernet was Internet of Things because it has a cable. And then I realized that the label lost any meaning. So if you have to not use digital twins as a label for anything to do with emerging tech and digitalization, right? For example, it is… 

I think it’s debatable whether a drone doing an inspection, just taking a shot of, I don’t know, a crane. Okay. It’s debatable whether it’s a digital twin. It is in the biggest sort of sense of the term, but it’s actually just taking a photo, right? So if you have to really, now that we looked wider in this conversation, if we have to bring it down, what is not a digital twin then? What is not a digital twin in your view? 

Is it about being smarter? Is it about the smartness in the digital twin and data analytics? Or is it just about capturing some information in the environment like people have been doing for a while? So is it about the intersection of sensing with AI, perhaps? 

 

Justin (26:30.057) 

That’s interesting. So, I mean, the first thing I’d say is that it’s not so much that if a drone takes a photograph, you know, you then have a digital twin. The drone would still need to provide that information back to a brain for it to process and create the virtual representation or digital representation of that crane. So the drone is a part of the system. And, you know, what has happened over recent years is that we have 

digital models that aren’t connected to the real world, but they can still be used to simulate different scenarios. Or you’ve got digital shadows that are connected with a one -way connection, pulling information out to the brain, and then the brain will work out what’s going on but doesn’t have that flow back again. And then you’ve got the twin that goes both ways. Now the reality is, despite the fact it’s been around for a long time, 

It’s very rare to see a fully operationalized real -time digital twin running of the type that Rolls -Royce has built into its engines in order to ensure information flow about the performance of the engines in real time. It’s very rare to see that day -to -day in large industrial scale assets, although we’re seeing some of that happening across the spectrum. So… 

you know, it’s still an aspiration to get to that fully operationalized digital twin running your operations where you can just stand back and let the system run itself. We still need that manual intervention. So, you know, whilst, yes, on one side it’s boring because it’s out there, I think it’s still aspirational. We still haven’t got there. But as a term, if you then look at it, you know, of course, you know, the marketing world shifts and moves and names. 

categories of technologies in different ways. One of the programs that I’m involved with, the joint program lead, is the National Cyber -Physical Infrastructure, which is another name for digital twins plus other technologies that sit around it. Perhaps in some ways it does help to address the challenge though that has been perhaps put forward that digital twins have become boring and now encompass sort of too much technology. 

 

Justin (28:56.293) 

Cyberphysical makes it clear that you’ve got this direct link between the real world and the digital world. Is it just a name change? Or are in fact there other technologies that sit inside cyberphysical infrastructure that we should be treating differently? For example, autonomous robots or autonomous vehicles of some sort. Is that a digital twin? 

Well, of course, on the side of it, I would start to argue that you’d need the digital twin for the robot to know its environment and work its way around it. But the core technology of that robot arguably sits in a different category than the twin itself. So what we see are these technologies, certainly in the digital world, that all sort of come together and require different types of technologies in order to do, ultimately, the job that we ask them to do, which is to solve. 

 

Maurizio Pilu (29:29.446) 

Mm -hmm. 

 

Justin (29:47.812) 

some of our real world problems. And whilst we’re describing the digital twin from the sort of roots up rather than the problem down, I think that these technologies need to work together. We need to understand how to bring them together, how to integrate them, how to share the data and get that data in the right place. And what we’ve seen is the rise of data sharing platforms as a way of being able to help manage that. 

authorize different users and authenticate different people and connect with different platforms and share them across different sectors. That’s a very difficult thing to do. And it’s one of the areas that we’ve focused on is how do you get different sectors with different regulators sitting behind those sectors to share their data for the greater common good. And often the challenges that sit there are around issues that are linked to 

the value realisation of doing that. Somebody might win more than somebody else by sharing something. But for us to take that next step, we need to recognise that the greater good will require us to regulate different industries in a way that encourages that or enforces that, if we can’t just encourage it, aligned with policies and programmes that ultimately are linked to legislation and funding. 

 

Maurizio Pilu (30:51.494) 

Yes. 

 

Maurizio Pilu (31:12.774) 

Thank you, Justin. So we are almost reaching the end. I wanted to ask you perhaps a final question. And I’m going back to that individual buyer. So you now delved into, I think, more of the policy, the grand scheme of things, which I think is what sometimes put off people like, whoa, it’s complicated. So I would like to go back to the boring. 

By boring, I mean practical. I can buy it. It does something for me right now. So that was the compliment there. Boring is good. Very ambitious means I need to wait 10 years for it to happen. So we know that there is a lot of practical solutions that can be called digital twin right now. And I think they solve real problems, like condition -based mentoring, predictive analytics, advanced inspection, and so on. 

My question is the following. Clearly, here is going to be a journey towards the grand vision. And like any technology, it will take a while before it’s adopted and industry figures out what to do. So it’s a journey. But we are now here, right now, with a lot of technologies that can do this out there. So going back to that pub, that hypothetical pub where you pitched, right? 

Let’s say your picture was very successful and somebody says, wow, I really love it. What do I do next? So just a couple of recommendations, I would say, for practitioners that would like to implement better sensing and better visibility of their assets to manage risk, improve efficiency. So what would you say they do next? 

 

Justin (33:07.583) 

Well, Maurizio, you talked about the journey and I think the journey is actually a better way of thinking about a digital twin because it’s not a thing, it is a journey. And the map that we follow on that journey, of course, is going to depend on a couple of key things. First of all, where are we now? And not all industries are at the same place. We see other industries moving ahead at a faster speed and some lagging behind. 

So depending on who, I think it’s important to understand where are we now. And that will be a combination of what technologies an organisation already has, what capabilities it has, its dependencies on being able to take things forward, how much is it dependent on the large systems integrators to help that journey. But I think it’s important that the skills and capabilities to do this reside. 

within the organisations and building those skills, that knowledge and that insight and understanding how it might apply to that particular industry and the particular key use cases and strategic direction of the overall corporate transformation that’s going on is key. So understanding where you are, being able to recognise the skills that you’ve got, the technology that you’ve got, the supply chain that you have around you to be able to support that. 

understanding where you’re going, what are the key use cases that will provide short -term value, not the 2050 future view of how the entire industry and sector will change, but what’s the next thing that you need to do to be able to continue on that journey that will allow you to maintain momentum, will allow you to build support. I’m assuming that the guys and girls that I’m talking to in the pub are interested in 

innovation want to drive some change as opposed to just looking as business as usual and try and mitigate risk by reducing the amount of innovation that comes into the operation because that creates risk itself. But let’s say we’re talking to innovators that want to drive some change, creating that roadmap, the journey from A to B, making sure that along the way the value is being realized at each step, allowing you to build a bigger and bigger… 

 

Justin (35:29.468) 

business case as you progress and the journey unfolds I think is key but thinking about it as a journey and what next and how you’re going to build it is you know the core thing to get I think a level of understanding around so you know that might be adding more sensors it might be… 

 

Maurizio Pilu (35:46.63) 

Thank you, Justin. No, no, thank you. We at SafetyTech Accelerator, we’re working with dozens of clients looking at this type of innovations right now, in fact. And we always advise, I can only second what you said, we always advise and we actually do develop roadmaps with them. And we always advise to… 

and think about the short term, medium term, long term, and then focus the actions on the near term, where they can get results now. So that is a great piece of advice that we can only second. And Justin, I’m afraid we have to wrap up now. I guess we could talk about this for a long time, you and I, but we need to wrap up. So I would like to thank you very much for joining us today and for the stimulating discussion. Thank you, Justin. 

And thanks everyone for watching this episode of our series. I would like to give particular thanks to our marketing team that made this particular episode happen. And if you would like to discuss these topics further or you would like to suggest topics for future episodes, please don’t hesitate to get in touch with us. This is Maurizio Piru from London and on behalf of all of us, goodbye.