(Not) Just Another Digital Twins Presentation

SEAN OLCOTT: Hi, my name is Sean Olcott. I'm from Gafcon Digital. This is our recorded presentation of Not Just Another Digital Twins Presentation for you.

I will be joined by a compadre Casey Mahon from Autodesk. He's going to present a little later. Then, my colleague, Hal will have presented at AU, but he wasn't able to attend our virtual recording. So you won't get to see him today.

Here's a Safe Harbor Statement. If you want to read it, you can press pause on the video. I'm just going to move forward.

All right, so as I mentioned Hal and I work for Gafcon Digital. Casey is with Autodesk. I've known all of them for a number of years, Casey going back at least seven or eight years from when I was working at Gafcon and he was still working for an architecture firm. And Hal, well, you won't get a chance to meet him, but he has spent about 30 years in the industry. Before he joined us, he was the Director of Innovation at Skanska in their southeastern division.

So real quick about the company that I work for, we are a systems integrator technology consulting firm. We used to be part of an owner's rep called Gafcon, which is how we got the name Gafcon Digital. In April of last year, we spun out our technology services group out of Gafcon, which is a top 50 third-party [INAUDIBLE] firm based in Southern California. And within less than a year, we were acquired by Anser Advisory, which is a top 15 [INAUDIBLE] firm in the country.

And then a couple of months later, they were acquired by Accenture. So this gives you an idea of roughly where we sit. Essentially within Anser, we're within their innovation group. And then within, Accenture, we're within Industry X, which is where their capital projects, capabilities, and expertise are.

But generally speaking, we are totally agnostic when it comes to the technology we use, the vendors we work with. We are focused on not just design and construction, but also the build op, the built asset operational lifecycle too. So a lot of our solutions leverage technologies and tools that are essentially implemented post construction.

So regarding digital twins, there has been-- there's a lot of digital twins presentations at the conference this year. And there has been for several years. It is a both an exciting, but also kind of confusing or complex topic in the sense that you have as a concept, there's many different enabling technologies. So depending on the technical domain of the person you're speaking to, they could be talking about digital twins and they could be talking about very different technologies from the person who's talking about digital twins in the next room over.

Additionally, there's also a number of related concepts, which digital twin overlaps. They are a part of everything from industrialized construction and sort of DFMA sibling, as well as any sort of smart solution from smart buildings, smart cities, smart countries, et cetera. A lot of overlaps with Industry 4.0 and the product lifecycle management generally. And then also the metaverse, a lot of the enabling technologies are shared between the two.

But what we're going to do for the purpose of this presentation is we're going to use some definitions that, at least, I'm very familiar with. We were a founding members of the Digital Twin Consortium, which was a cross-industry consortium made up of a number of notable industry companies and academic institutions to focus on better kind of clarifying the digital twin as a concept and ensuring that best practices were documented and communicated to the industry.

I ran our subcommittee in charge of glossary or the terminology. One of our main deliverables was a glossary of terms, which was a excruciating process of herding multiple cats who all had strong opinions as to what terms should be included and specifically what the definition should be. So I'm going to use these definitions, because I bled over the creation of them. It will also clarify some of the terms we use later in the presentation or recall back to.

At a high level, if you think of a digital twin, it's just a virtualization of a real-world thing. There is a process of synchronizing it with [AUDIO OUT] at some frequency or fidelity that's suited for a particular use case. It was not a term in use at the time, but a very kind of notable example of it being implemented was by Rolls-Royce and their manufacturing group that provided aircraft engines, in which they essentially completely transformed their business model and focus to providing engines, which essentially got-- they were being paid for the use of the engines and they were no longer being purchased by the airlines.

The concept of a digital twin system, we'll refer to a lot, which is imagine there's a system of systems that allow you to implement a digital twin for some particular use case. If you think of when you ordered an Uber, that is a digital twin system in which there's multiple digital twins of the cars, the riders, as well as digital twins of the computational models that calculate how long it's going to take for them to pick you up. And it's all leveraging a mobile phone as a platform for delivering it, so that's a digital twin system.

And then the third one is a term that there is-- we're not going to go into semantic discussion and distinctions around all-- there's many kind of versions of this definition. This is the one-- I'm going to be using it for this purpose, which is the concept of a digital thread. And a digital thread is essentially that mechanism that ensures that when you have lots of different models and you are using data across those models and across time for some particular use case that it remains consistent.

And so if you are using Google Street View, you not only see where you're standing on a map, you can also see the 360 photogrammetry, UI in front of you, as well as, depending if there's other information about shops or things located to you, you're also seeing that information surfaced. That all relies on a digital thread to make it possible. So these are terms that will be that essentially you experience or use-- or experience in some form or fashion every day now with other applications and systems you're already using.

So the reason why it's notable is because it's getting easier. It's easier to get digital twins. And there is essentially more and more value to using them. You got more computers than you ever had, they outnumber humans. The algorithms are getting better. We're seeing they don't even rely on humans entirely anymore.

There's more access to computing power. Witness Moore's law. There's also-- we're more connected than we ever have been before in terms of-- in the cloud, at the edge.

And then as you get more and more data, there's a sort of infinity loop of improvement that's possible. So it's an exciting opportunity in terms of what's now possible. And it's becoming more and more impossible and even easier to implement in sort of where we are in this current stage.

So there's two kind of categories of virtual representation if we're going back to the definition. And this is an important one. Because depending on who you're speaking to, it could be-- they could be talking about model in a different way. And so there's virtualization. There's kind of two categories. You've got the stored model, which is kind of structured information. It's usually representing the state of some entity or process.

And then, the computational simulation model is basically-- it's what you would do for simulating how long it's going to take your car to get from your home to the McDonald's. So there's a data and algorithm that you input data in, and then outputs out to another stored model. So these digital models depending on who you talk to, they could be a mathematical equation or it could be a point cloud.

When we refer to entities, entities are basically anything essentially in the real world. It could be weather. It could be a vehicle. It could be a building. It could be a space in a building.

It could be a system-- a building system. It could be a piece of equipment within that system or subcomponent of that particular piece of equipment. And you could have stored or computational models related to that entity. And then when we talk about processes, these entities essentially combine and interact to form these systems in which you have some processes being done where it could be a productive process in which something is being made or it could be non-productive like your ice cream melting in the back of an ice cream truck when it's not running.

And then, the notion of synchronization is an important one. Because absent synchronization, you're not really twinning anything in reality. And so there's a bidirectional element to it, where you can imagine-- you can see, for instance, something that's happening in the real world is being essentially replicated in its virtualization, so you can see it. But there's also the control. If any of you guys have had any experience with building automation systems, there is that sort of virtual-to-real capability for digital twin use cases.

Another one is the idea that you don't necessarily have to have a robot providing-- doing the work for you in terms of creating the synchronization. That mechanism sometimes can rely on a human. If you think about just the process of going out and doing a laser scan of a building, that in effect is a human in the loop. Or, someone going out and fixing something in the real world and turning a switch on or off, that essentially is essentially an example of a human in the loop. So it doesn't have to be exclusive entirely to robots.

And then, the last one is the concept of fidelity and frequency. And so we use these terms. Fidelity is essentially how accurate or faithful it is to the real world. And then, the frequency is how often. So depending on your use case, your fidelity or frequency might be different.

If you are going into Zillow having-- doing a virtual tour of a house that you're interested in, that doesn't need to be synchronized very often. And it doesn't have to be super high fidelity, because what you're just trying to do is get an experience for walking through. But if you're doing job walks or you're going to have some trade come in to do some installation, you're going to want to do a laser scan. You want a LiDAR model to more accurately represent it. Or, if you're tracking progress of stuff being implemented or installed over the life of a project, you may have the cadence or the frequency by which you do the synchronization may change.

So if you were going to apply criteria to what a digital twin is, essentially, what you're saying, is there are some virtualization? Is there a synchronization? How is it visualizing, or simulating, or allowing you to control this real-world entity or process of interest? And then, ultimately, if you're evaluating or prioritizing these, the benefit of a use case, you have to understand, well, how is it making something better? How does it represent some transformational way of managing or making decisions related to this particular real-world entity or process?

Now, there's a lot of-- obviously, because it's a popular term, and it's generating a lot of momentum within the industry, there's many, many solution providers working within the AC space or talk about digital twins as part of their general solution offering. One category we use is professional services, a sort of generalist one. So if you think about a company like Deloitte, or EY, or Accenture, they are essentially-- they're doing digital transformation initiatives for big organizations. And digital twin is effectively sort a subset of that.

Professional services within the industry, I would characterize these as your CBRE, or Anser Advisory, or even a lot of [INAUDIBLE] business is this, in which, is essentially, any digital twin solution is really-- it's just an add on to the core solution they provide as an outsourced partner for managing the delivery or management of a built asset. Architecture and engineering firm-- or [INAUDIBLE] service industry tech, like if you think about your-- kind of the people who traditionally have been channel partners or service providers focused on a particular vendor or category of technology, they're doing digital twins too.

It's usually tied to their domain of expertise. So they have been a GIS-focused technology implementation partner to Esri or some other company, they would be focused primarily on GIS-based use cases. Whereas, a lot of the CAD resellers would be focused on BIM or Reality Capture-type ones.

Architecture and engineering firms are also talking a lot about digital twins. And again, that's usually an added service. And it's typically essentially part of their traditional kind of design engagement and they're able to add another line item to it. And similar category for general contractors and trades. Although, there is a distinction because, sometimes, the contractors are providing post construction services if it's like performance contracting or things like that.

And then you have, of course, your software vendors who have the providing-- are providing the platforms and applications that would make up the subsystems of a digital twin system. That includes Autodesk, but also includes many others. And then, the hardware vendors who are essentially providing that hardware that allows for that synchronization with the digital and physical twin.

What you'll find is that increasingly there's all the ones who don't do technology or that's not their core kind of area of focus are moving into technology, and then the ones which are centered on technology are increasingly centering a lot of their effort on building out software. And so you see a lot more custom digital twin software solutions on the market. Some of them are very kind of narrow, very much point solutions. But some of them are being provided as sort of added services from a general contractor or a professional services firm.

The focus of the conversation that we would typically recommend is on-- not what is a digital twin or even the type of technology, but really, what type of-- where there's value in transforming the process by leveraging this concept of virtualizing some real world thing. So if there is going to be a discussion around what you should implement, it shouldn't be because you've seen a really cool video of which some sort of amazing capability has been shown, but has not been demonstrated to be scalable or to deliver substantive value.

It should be really on, well, where are the use cases that we have the most value? What are the processes that we should be focusing on? And what are the outcomes that we want to achieve? And will this help us achieve?

In terms of when you can have a digital twin or have a use case of a digital twin, it's clearly a use case that at the completion of a construction project, if you organize all the data, and artifacts, and collateral, so that it can be integrated in some operational systems. That counts as clearly as a use case. But given the definition that we're applying to it, we think there are many times prior to that or-- for other purposes in which you can have a digital twin use case.

During capital planning, there's a significant amount of work that goes into understanding what the requirements-- or what you should actually build, or program, or design. A use case of digital twin during capital planning might have actual an IoT data historian about how space was actually utilized and to help that kind of drive your kind of-- the programming decisions you make when you're kind of before design is started.

Another one might be during design. And this is a fairly-- again, this doesn't have to be an interesting use case to be a use case, where you could just be grabbing LiDAR models and integrating them into your BIM authoring process. So that what you're working within, especially if you're doing a fit out of a complex space where there's not a lot of room for things, having that high fidelity representation of that real-world constraint, that counts as a digital twin too.

During construction, there's a significant amount of progress being made in terms of incorporating both IoT and reality capture data to document work that's actually been done. And so there's some really interesting use cases, or possibilities, and benefits around of streamlining or improving the pay-out process by incorporating that type of data. Again, that's the digital twin of a construction process.

And then during closeout, when you can confirm or just verify that what's actually-- what you're actually providing aligns to both what's been documented, but also what's been designed. Those are all use cases. They start before design, and they finish at closeout. So when you are talking about a digital twin in this use case, there are products that are being sold, which have a digital twin use case.

But it might not be the only time you're using a use case. So our general perspective is to be more broad and accepting in terms of what you would call a digital twin and be more focused on, does it satisfy your digital twin use case? And is it an interesting use case, or a compelling, or beneficial one.

So the way we approach this, as a systems integrator, is to take a composable approach, in which, depending on the client's requirements, they have some problem that they want to solve that cannot be either-- cannot be addressed by their current ecosystem, or even off-the-shelf products, or could be more-- could be solved in a more effective way or scalable way if it wasn't reliant on a narrow-point solution, or added licenses, or added capabilities for an existing system. So typically, what we're focusing on is those type of things that are not easily solved by their existing [INAUDIBLE] or their existing project management system or other systems.

And so within that approach, you have a process of determining what models are needed for the digital twin. It would again be driven by the use cases you have in mind. How those are-- how the various kind of a data sources are integrated, and then implemented into some sort of data management infrastructure to create your digital thread, is an element that is required.

And then, for how you present or make that data available to support some sort of the workflows needed for your digital twin use cases, that design process-- they're all interrelated. And depending on the requirements, that drives, essentially, what the platforms that we would choose for the job would be. In some cases, you may have a use case in which it's a geospatial platform is much better suited for how you would want to present or contextualize that data.

Whereas, in some cases, you would want a sort of a model-based or 3D-based UI. In some cases, you might want a 2D-based UI or you might want a knowledge graph. All of those might change both how you design the underlying data management infrastructure and implement it, and then how you design and configure this presentation layer. So now, I'm going to pass it over to one of our partners that we work a lot with-- Casey Mahon over at Autodesk.

CASEY MAHON: Thanks, Sean. As Sean mentioned earlier in the presentation, my name is Casey Mahon. I am with Autodesk Tandem.

But prior to that, Sean and I-- I've been trying to get Sean to collaborate with me for about the last 8 to 10 years, so for as long as I've known him. And as he stepped into a new role at Gafcon Digital and I stepped into a new role here at Autodesk Tandem, we found a really strong opportunity to do that. And mostly, that's building off the framework that Sean just spoke to and the opportunity that Tandem presents.

So Tandem, as a product, was created to support digital transformation in empowering our customers and partners, like Gafcon Digital, to realize their digital twin visions. We do that by focusing on twin building, which provides repeatable workflows for defining the building's digital twin, harnessing BIM data to create and maintain the data model, ensuring completeness and accuracy, and accelerating operational readiness, and then supporting insightful operations by focusing on facility monitoring that looks to drive facility operations professionals, provides insightful information to help inform better decisions, and ultimately, to improve efficiency and reduce operating costs.

We do that in a number of ways and through a number of leveraged features and capabilities inside of Tandem. Both defining assets, the systems of systems, that Sean mentioned earlier, spaces, streams, functionality, and looking to the future to tag data feeds and behavioral impacts all configurable and conformable to support the use cases defined earlier in the twin building process.

So we really see that combination of use case definition capabilities within the product to start to stack up on each other to enable these insightful operations to visualize and contextualize data in space about systems and objects to improve performance based on actionable insights, and to measure performance increases or reductions depending on the use case opportunity that was defined earlier and will benefit the customer. And, Sean, back to you.

SEAN OLCOTT: Thanks, Casey. Well, if you had been in person at our session in AU, you would have seen Hal step up now and do a live demo of what I'm about to talk about. Obviously, you can find us, our information. So if you do want a live demo, just reach out.

But what I wanted to talk about specifically was the Smart Jobsite. We have been doing a lot of work with IoT systems for a while, Gafcon Digital. And initially, a lot of our effort was on post-construction use cases, so a lot of smart building-related or smart factory-related implementations.

However, there is an increasing demand for IoT systems in real estate and construction. There's a lot of compelling scenarios where you'd want to have them on a construction job site. They're very dynamic environment. There's lots of things changing.

There's safety. There's risk. There's quality concerns. And so we began doing some exploration around what a digital twin of a construction-related process would look like leveraging AWS or Azure IoT stack and their respective digital twin platforms. AWS has one called IoT Twinmaker. Microsoft has Azure Digital Twins.

So initially, we built this up. We have a partners in place with hardware-- partnerships in place with hardware vendors and kind of a proved set of telemetry devices that we could install or implement on a construction job site-- for owners, we work for owners. And what we found was that-- it's very powerful to be leveraging those systems, because the existing IoT systems being-- solutions being used on construction projects are typically just deployed by contractors. And they're usually point solutions. They're very narrowly-focused stovepipe solutions, which go from basically hardware to insights focused on a certain area.

But our perspective was that the hardware should be essentially fungible and that you should be able to swap it out depending on the type of telemetry you wanted to be tracking on a job site based on your particular requirements. And if you use AWS or Azure, you can integrate those systems relatively easily and surface them in much more interesting and informative ways that correlates information across different IoT systems. So we did that. We actually have working prototypes and some demo implementations in place for both IoT Twinmaker and Azure Digital Twins and their respective IoT stacks separately from one another.

But what we found was that it really would be more compelling from a visualization or contextualization standpoint if we could use the platform, which was better suited to engineering or construction-related data. And so we had these systems, which did some monitoring of the environmental conditions at the job site for safety purposes. Some of our clients, we have some data center clients where they have to maintain the integrity-- operational integrity of existing operations while construction projects are underway.

That's a particularly big issue within health care, as well as data centers and other types of facilities where the projects are going to go on and operations are going to go on regardless of whether there's construction activity next to it. And you have to be able to essentially track, and monitor, and even manage an audit trail of critical site activity, as well as introduce some kind of monitoring of site access and security that doesn't have to be done in a sort of siloed stovepipe solution.

So what we thought was exciting about Tandem is that you already have a very interesting and compelling upstream pipeline into data that's used during the design construction process, as well as a sort of a UI and workflows, which makes sense for a construction activity. So we have within the Smart Jobsite, we have-- there's more telemetry that we can track. We found that there's a handful of high priority ones that we want to be doing.

One is a pressure, air quality, noise, water consumption, and power consumption as well as security. And this is an example of very high level abstract diagram of how this works within our stack. With Microsoft tools, you could swap out the Power BI logo with Amazon's Managed Services Grafana. BI tool, you could swap out Azure Synapse with AWS's IoT stack and an associated data management systems.

And then, you could swap out Azure Digital Twins for IoT Twinmaker. So this it is a deliberately composable solution that depending on our client's preference for AWS or Azure. We could go either way.

In some cases, we actually do have an implementation, which uses both. But that's for a live-operating facility. But this is an example where you can see heat maps according to space, but you can also track information and telemetry at an asset level and not just in a room level.

There is some challenges when it comes to addressing needs around tracking assets or critical equipment within Tandem. They don't have real-time location services yet. Although, there are some interesting methods or ways that you can manage or mitigate that.

But for now, we've been able to-- really, really pleased how easily or how good the pipelines are from Azure's IoT stack into Tandem, where we can still use Azure and leverage Azure's system for being the IoT data historian, a lot of the kind of data management, some of the deeper analytics if we want to access some of the services. But from in terms of surfacing or contextualizing a real-time data, very, very excited about what Tandem can provide.

So before you get to operations, before you've collected and organized all the data and artifacts for handover, before you started looking at smart building implementation, there's use cases for Tandem as an IoT systems integration tool and platform before you get to operations.

So had you been at the in-person meeting, you could have seen this being demoed in action. You know where to-- I'm easy to find. Reach out to me if you did want to talk some more about this.

Casey, thank you so much for participating. I wanted to make sure you had a chance to talk a little about Tandem. And I hope you all enjoyed the presentation. Thank you.