Finding Success With Digital Twins

This video was sponsored by ANSYS.

Learn more about ANSYS Twin Builder and start a free trial at ANSYS.com.

The following transcript has been edited for clarity.

Michael Alba: Hey everybody, and welcome to Designing the Future. Today, we're going to dig deeper into digital twins to discover some successful implementations and typical pathways to success. We're joined by Manzoor Tiwana, lead product manager at simulation company, Ansys. Manzoor oversees Ansys Twin Builder, the company's product for creating digital twins, and he's previously held positions at Autodesk, MathWorks, and Bosch. He's got an MBA from Carnegie Mellon, a master's in automotive engineering from FHT Esslingen, and a bachelor's in mechanical engineering from UET Lahore. Manzoor, thanks for coming on the show today.

Manzoor Tiwana: Thanks for having me.

Michael Alba: So, you've been in the industry long enough, I'm guessing to have seen maybe a few different interpretations of the digital twin. Could we just start today by going back to basics? Could you tell me how does Ansys define the digital twin today?

Manzoor Tiwana: Yeah. As you said, the term digital twin has been used in a lot of different contexts in the industry. And Ansys is part of the Digital Twin Consortium, and we have adopted the definition of Digital Twin Consortium. And they define digital twin as a virtual representation of real-world entities and processes, synchronized as a specific frequency and fidelity. What that means is that on one hand you have your physical asset and on the other hand you have your model or the visual replica or digital twin, and you want to connect them with some sensor. And with the help of this model, you can track the past, what happened to your asset. You can generate deeper insight into present, how is your asset is performing. And you can also predict and influence future behavior.

The thing that is very unique to Ansys is that it allows you to build these models from both physics or simulation-based model, what we call them, and we can also add data analytics to create those unique insight into your operations.

Michael Alba: So, what are some typical applications you've seen from your customers?

Manzoor Tiwana:  We have seen a large-scale adoption in several industries, from automotive to automation. But there are three major areas or applications that we have seen a lot of traction. Number one is this industrial flow network. This includes like the performance of your overall fluid networks, performance of your rotating machines, like pumps, compressors, turbines, process optimization of fluid mixing and blending processes.

The second area where we have seen a lot of use or a lot of traction is in the automation industry or in the elective drives. Some of the typical application includes the performance and thermal management of the drives. Also, in the EV and HEV space, in the automotive space, we have seen the thermal management of batteries, battery packs, and also the complete EV powertrain. So, these are all the application we have seen in this area.

And there's this new or emerging technology or application what we are seeing in the industry, is also the heating and cooling application, like optimization of HVAC systems, zero carbon, and carbon capture. These are all the application what we are seeing in the industry.

Michael Alba: So, wide-ranging. Are there any particular exemplary success stories you've seen with any of your partners that would illustrate this concept in a little more detail?

Manzoor Tiwana: Sure. We have seen several customers from, as I said, we have seen a large-scale adoption from automation to automotive, and we cannot talk about all of them, as the customer relationship or they have not signed those disclosure agreements with us. But there are some customers that have done a press release with us, is ABB, EDF, A123, Volkswagen, to name a few.

One of the recent or the newer story is the ENGIE. They are the world's leading supplier of energy efficiency services, and they are helping their customers to transition from traditional carbon-based energy to carbon-free energy. And they are employing digital twin to reach that goal, to improve or optimize the burning process to reach a zero-carbon target. In addition to that, we are also working with a number of partners in the industry, like Microsoft, Rockwell Automation, to deliver the solution to our customer.

Michael Alba: So just on the topic of those partners, you are one of the founding members of the Digital Twin Consortium, which I believe began early last year, could you update us on the status of the Digital Twin Consortium and Ansys's role in it?

Manzoor Tiwana: Ansys is the founding member of the Digital Twin Consortium, along with those companies like Microsoft, Dell, and GE. And the goal is to drive the development and adoption of digital twin technologies, also to drive common terminologies and standard. One example is the development of digital twins definition language or DTDL. So, Ansys is collaborating with Microsoft to work on this DTDL. And we are also working with Microsoft to create the reusable reference architectures. So the goal is or the aim is to help these IoT solution to drive this standardization and help these IoT solution to talk to each other from multiple solutions, so you can combine them in a single solution.

Michael Alba: And how far along is this project? I mean, how mature is digital twin technology at this point? Are there still a lot of technological obstacles to overcome before we can really get to the vision that you've outlined?

Manzoor Tiwana: So digital twin technologies, I wouldn't call them as emerging technologies anymore, so we have used this term that it is emerging, but I think it has established itself quite a little bit in this area. Having said that, several companies are still on the journey to explore and adopt the digital twins. One of the challenges that our companies are facing is both people that the organization need to evolve and adopt this technology, and also on the technical side, some of the typical challenges they face is how they want to put the sensors inside, how they want to collect the data, how they want to model these digital twins, from where they create those insights.

So, these are the very typical challenges that especially if you are basing your digital twin or model on the analytics only, it requires you to collect a large volume of process data for training and also the accuracy might be insufficient and limited just to the observed data and to the available sensors. On the other hand, Ansys provide physics-based models and virtual sensors that can integrate your physics-based model and the data analytics together to generate better understanding of your operations.

Michael Alba: Could you elaborate on that concept of virtual sensors? What does that mean?

Manzoor Tiwana: So virtual sensor is, in the physics-based modeling, you can predict quantities that are not measured directly from the sensors. To help understand you, I'll use the example of an electric motor. For example, you have an electric motor and you want to, say, find out what is the temperature of your rotor is or what is the temperature inside the motor is, the one way to do is that to put a physical sensor inside, but that is not all feasible or it's not very cost effective, or sometimes it's physically not even possible to put a sensor inside. So, with the help of other sensors, for example, with the help of current, you can predict what is the temperature inside of your rotor is, so you know the properties of the material of your motor.

And with the help of some sensors, you can predict other sensors or other quantities like temperature in this case. So if you have increase in load, so that will increase the current in the motor, and with the help of these models, you can predict what's the temperature is going to be in the future, what's the temperature is currently. And also, if you keep on using this motor, how long you have till it will reach the critical temperature. So all these kind of things you can do with the physics or the virtual sensors.

Michael Alba: So, in your digital and physical twin, you've got a combination of physical and digital sensors, giving you a complete picture of the system. What's the balance between those two types of sensors. Could you theoretically have all physical sensors and still make use of a digital twin?

Manzoor Tiwana: Of course, you can have both, all your data, it could come from your physical sensors, but as I said before, it's not always possible to have those sensors there. And it's usually not very cost effective way to put a lot of sensors inside. One example is that some of our customer, they have these brownfield applications, where they have these large applications where they have built 10 years, 20 years ago, and they want to now employ digital twins there. And to do this, they have to put new sensors inside. So the cost of putting these sensors inside is really great, it's prohibitive. And with the help of some sensors, like, say, you can put some sensor as a flow sensor, and with the help of those sensor and using these digital twins, you can predict how your system is going to behave. So you can use a subset of sensors, but still get the full accuracy.

Michael Alba: So, you're talking about essentially simplifying the complex system and this is an important point, I think you bring it up often when talking about Twin Builder is this concept of model order reduction. Can you tell us about that and how it's achieved in Twin Builder?

Manzoor Tiwana: So a reduced order model or a ROM, how we call it, it's a simplification of 3D physics. As you know, the Ansys is the leading authority in the simulation of 3D physics. So if you have a 3D physics model and you want to reuse that in a system simulation perspective, what you can do is that you can simplify that model using this reduced order modeling technology, without losing essential accuracy. Most of these 3D physics, it takes a lot of time to simulate, so which that is not very feasible if you want to generate these insights in the real time or near real time perspective. So you can use these 3D physics model, you can simplify them without losing those accuracy. And you can use them in real time or near real time for generating those insights.

Michael Alba: And is that a process of reduction something that I would do in Twin Builder or do I reduce my model and then bring that into Twin Builder? Could you take me a bit about how it works that way?

Manzoor Tiwana: Yes, sure. So the process of generating these reduced order models is in the way that what you have to do, you can use either Ansys physics, but it's not limited to Ansys physics, you can also use any third party simulation software. And you generate inputs and outputs, and you generate that data, how your system is behaving. And then you bring that data inside Twin Builder, and then you can create those reduced order model inside Twin Builder. But once you have created those model inside Twin Builder, you can export them as a twin, what we call a twin file, and you can then deploy those reduced order models with other system simulation capabilities. And then you can deploy them on the cloud or on the edge or on any IoT platform.

Michael Alba: What is a typical journey for your customers in order to implement a digital twin, to get started from zero, what do they usually do?

Manzoor Tiwana: We recommend a crawl-walk-run approach to our customer. The very first step is to identify an application that is a low-hanging fruit. That means that it has some economic impact, so it could be in two ways, either the asset itself is very expensive, so any wear or damage, it can be very costly. For example, a wind turbine, right, so wind turbine is extremely costly. And any damage or wear happened to that can be really expensive. Or a turbine, a water turbine, they are some few examples where the asset could be very costly and any wear or damage to that asset could be very costly. Another aspect or economic impact could be that the asset might not be very expensive, but the downtime is very costly. For example, if you have a pump that is pumping oil from undersea, if that fails, the economic impact of the downtime that is very huge.

The asset might be few hundred thousand dollars, I don't know, but the downtime could be millions of dollar. So first step is to identify an asset or process that has an economic impact, a large economic impact either by the downtime or in terms of its own cost. And the second step is to do a quick POC. That means to work with Ansys to create a quick proof of concept that takes four to six weeks. And after this proof of concept, present that to management and also gather the lesson learned. And in the last, we would recommend that after those lesson learned, you scale that to other applications, to the whole department or other department in the company.

Michael Alba: Now, when you talk about this low-hanging fruit that the company should start with and you look broadly around at all the possible applications of digital twins, is there anything you see as really ripe for digital twins that nobody seems to have really taken advantage of yet?

Manzoor Tiwana: As I said, we have seen a lot of application in all the industries. One application which is catching my attention or I have seen a lot of adoption in the industry is around heating and cooling application, like the refrigeration, HVAC application, zero carbon and carbon capture. Navantia is one of our customer that have adopted this simulation-based digital twin that help them monitor the performance and also manage the maintenance of their HVAC system on big ships. And with the help of this digital twin, they're increasing their productivity, they are reducing the downtime, and it's also helped them with the maintenance of their equipment.

Michael Alba: Now Ansys launched Twin Builder back in 2018, so it's been live for a few years now, and I'm sure you've refined it over those years. Could you tell us how Twin Builder has evolved since it's launch and what lessons you've learned along the way?

Manzoor Tiwana: Yeah, so Ansys has, as you said, we launched in 2018 and Ansys has established its position as leader in this space in the past three, four years. One key aspect of our go-to-market was to establish a key partnership with the IoT leaders in the industry, the people like Microsoft, PTC, SAP, Rockwell Automation, to build those relationship with the infrastructure provider in terms of cloud infrastructure or asset infrastructure and develop those connector, those out of the box implementation that user can take these digital twins and deploy on the IoT platform what they already have invested in.

We are also working to develop those design patterns and best practices. Another thing what we have evolved to adopt is the data analytics. So in the upcoming release, we are calling this the hybrid data analytics or hybrid analytics that allows you to combine physics with the data. That means that you can get the best of the both worlds. With the help of these simulation models, you can combine the test results and you can tune and calibrate your model to match the outputs to the real world outputs.

Michael Alba: Manzoor, I have one more question for you, and I'm going to ask you to put on your future looking hat for this one. But if you look into the future with this concept of digital twins in hand, how far do you think it's going to go? For instance, if I buy a car 10 years from now, am I going to get a file as well that has a digital twin of that car that I can keep an eye on or a house or something like that? Will digital twins start to accompany most or all of the physical things that we make in the world, in your opinion?

Manzoor Tiwana: The immediate application or immediate, as I said, the low-hanging fruit, is on the side of, with the things which have greater economic impact. For example, the asset is expensive, downtime is expensive, so those are the low-hanging fruits or those are the things which are first to adopt these technologies. But as we are going to evolve, also, as you mentioned, the cars, as the cars going to evolve to become more EV centered, more getting the batteries inside, so you have to monitor those batteries, you have to monitor how is your asset is performing. So as we going to evolve, all these technologies is going to be infiltrate also into the secondary application, secondary assets. And you will see that you will have digital twins of almost everything of interest, so you can monitor how your asset is performing in the field.

Michael Alba: And just a quick add on to that. What about people? I know Ansys isn't focusing on this, but some companies have explored this area. Do you think there'll be a digital twin of you and I, at some point in the future?

Manzoor Tiwana: So actually, we do explore these kinds of things, where we have some of the medical applications of the heart model and things like that, where we are on the experimental side, we are working with some bio companies too, to explore this, so to model how your heart is pumping, to model how your arteries are functioning. So all these things could be very, very interesting, in the future, before a surgeon can do a surgery, so he can do a surgery on the digital twin side. So it's still in the future, but I'm looking forward to that.

Michael Alba: Me too. I can't wait to meet my digital twin someday. Manzoor, thanks so much for coming on the show. It was great speaking with you today.

Manzoor Tiwana: Yeah. Thanks very much for having me. It was really great talking to you.

Michael Alba: And thanks to you for tuning in. We'll see you next time.