Digital Twin Real-Time Analysis

Hello everyone, welcome to a new video from Ozen Engineering. Today I'll be speaking about Digital Twin Real-Time Analysis. My name is Ahmed Elghandour. I'm an application engineer at Ozen Engineering.

I have a bachelor's in mechanical engineering from Caledon University, Egypt, and a master's and Ph.D. from the University of Illinois in Chicago. I have over 10 years of experience using finite elements with a majority weight on ANSYS products.

I have experience with structural analysis, linear and non-linear analysis, static and dynamic thermal analysis, and more. In this presentation, I'll be presenting the work I did with contributions from my colleagues Chris Cowan, Mesam Amiravran, Mohamed Zamek, Stephen Anshonglio, and Vahid Merafta.

A little bit about our company, Ozen Engineering. Ozen Engineering is an elite and ANSYS elite channel partner and distributor in California. Our work is focused entirely on simulation. We work with the ANSYS suite to help customers develop products successfully.

We provide software tools and engineering consulting services using finite element analysis, computational dynamics, electromagnetics, and other tools.

In this video, I'll be talking about concepts of digital twin and demonstrating digital twin prototypes we created at Ozen Engineering, including high-frequency and low-frequency models.

I'll also demonstrate digital twin IIoT or industrial internet of things deployment prototypes, including fluids application and multi-physics application. So, what is a digital twin? By definition, it's a connected virtual replica of an in-service physical asset.

When we speak about a replica, it's a virtual replica simulation. ANSYS technology enables digital twin 3D physics solvers, reduced order modeling, and twin builder platform. We use all these tools to create a replica of a physical model, which is usually in service, by connecting the concept.

The concept is based on the ability to capture and process data with low latency. We have high-frequency when we architect using cyber-physical systems, ML, using cybertran2 data tracking and technology. We can now do 3D enemies storage in a single 2D environment.

You'll import any data because additional software might clearly mention there are devs after the hand Guerrilla team.

We used per reminder mediumiséeons, TSP2, even course timetable processing, but mainly we are working hand over hand with Hai sir and handle big data, enabled through industrial internet of things, IoT, featuring connected devices.

It also enables system design and optimization, helps with predictive maintenance, and optimizes industrial assets management. So, what are the key capabilities of simulation-based digital twin? A simulation model and industrial internet of things and data analytics can be deployed together.

We can have the ability to assist in asset monitoring, predictive maintenance, and what-if analysis before applying the solution. The key benefits of simulation-based digital twin include increasing top-line revenue, managing bottom-line costs, gaining and retaining a competitive edge.

Digital twin is an interesting new technology, but the key benefits are financial. Digital twins offer value through both cost savings and new revenue.

The rate of interest or ROI comes from one cost savings related to warranty and operation to new revenue resulting from innovative business models and new products. Digital twin technology should be applied mainly if we have expensive to replace or expensive if it fails unexpectedly.

When failure modes cannot be reliably predicted with historical or IoT only data. The twin builder architecture and workflow are as follows. We create the simulation using the twin builder workflow and then validate the physics to make sure that the physics are working fine.

Then, using ANSYS Twin Builder, we will create the model and then create what's called a reduced order model (ROM) using twin builders.

All of these data will be used and then we can use this to couple it with one of the digital twin partners for the deployment, either PTC or Rockwell or Microsoft, etc. A reduced-order model (ROM) is a simplified and condensed high-fidelity model that supports a range of ANSYS physics.

It reduces solve time by multiple orders of magnitude, similar to response surface methodology. This enables real-time savings and offers the ability to expand usage of 3D simulation models to users who work at the system level. Now, let's see a step-by-step demo of how we digitize twin.

We need a little bit of flavor in here right now. So, let's have a look. Technology enables special multi-axis. Basically, we have an antenna that produces arrays, and we have a moving object. Based on these arrays and the object, we can always trace where the object is in an XYZ location.

We created a model using HFSS simulation, created the ROM for that, for the antenna array, and then went through digital twin.

We can see this is the output of the simulation based on the phenythrombin simulation, and then start to deploy it and compare real-life data, the output of the simulation to the HFSS simulation. And the real location of the car. This is a physical model he created.

You can see this is a target laptop to monitor and then location sensors and listeners to represent that. In this video, you can see a demo of what's happening. Here is a car moving in this space, ground.

You can see on the right and the left, we have the antennas that define the location of the car. At the real time, we can see the location of the car in the domain predefined.

Another representation here, you can see the car moving closer to one of the antennas, moving far from the antenna, based on how the car moves in this domain that we saw. Okay, another demo. Okay. Another demo.

It's a low-frequency electromagnetic field that's created by Michael Stevens and Daniel Esmaili. Basically, we have a coil that will provide a DC current through this coil and get a magnetic field aligned with the axis of the coil. Here is the coil. This is a simulation created using Maxwell.

And then based on Maxwell, we create a rotation. And then we can see that the current is going to be in the direction of the car. And then we can see the current moving at the direction of the coil. To all the current moving in this coil, what are we doing?

Well, basically, the bolts are out, and the thing will be good. So, what do we see here? What do we see inside a twin builder? We need to bring the room, track the output parameter that will be coming from the ROM, and then we compare that with a sensor that measures the real-life magnetic field.

To do this, we have a python code that imports live data from the power supply to be used as an input to the ROM inside the Twin Builder. Here's an example, the current changing, and the magnetic field changing with live data coming from the power supply based on the python code.

And we see the change in the magnetic field happening. I don't have a video for that, but this is what happened. The field will change live, and if we have the sensor in front of us, we can see a really good agreement between both of them.

Here is the workflow, the power supply provides current to the system, this is the physical system, and then to the data collector, and then using the python code provided to the digital twin, and then we can see the results coming from the sensor and coming from the ROM, which have a really good agreement in reality.

Okay, so now we'll show other prototypes that were created at Ozen Engineering that demonstrate digital twin IoT deployment. The first one I created this one with my colleague Steven Lu.

This is when we started to work on that, we decided to have something familiar to the area, so nothing is more famous than the Golden Gate Bridge. So for that, we created a digital twin IoT. This is one of the first prototypes that were created at Ozen Engineering that demonstrates digital twin IoT.

So for that, we created a CFD computational fluid dynamics model using ANSYS Fluent and then we can see here, this is some figures from the model.

We created the model on a fluid and then we created the ROM inside ANSYS Workbench and then to read the live data, we have a python code that reads the live data of the air computer. This is the signal that is coming through the bridge.

Basically, the main physics we are trying to do here is to predict the pressure on the walls of the bridge when there is wind coming in different directions. So based on the variation on the wind direction and velocity, what will be the change in the pressure on the bridge surfaces?

Okay, that's the end of the text. I've corrected all the spelling, grammar, punctuation, and formatting errors to the best of my ability.