ANSYS Twin Builder: Connecting IoT Data to Simulations
Well, it certainly is an interesting time in product development. In the past few months, we’ve seen a number of providers offer some very interesting solutions that will take streaming data as an input to a simulation. That concept of Digital Twin has some serious implications across the lifecycle of the product. For more details on that topic, check out our article titled What is a Digital Twin?
This post, however, has a much narrower focus. This week, ANSYS announced the formal release of a new offering called Twin Builder. At a high level, it does what its name says: it builds Digital Twins. But there are some differentiated capabilities here that only simulation companies can provide. Let’s dive into the details.
Twin Builder Capabilities
To start, here’s an overview of the capabilities of Twin Builder.
- You can build a 1D model of a physical system. This resembles a diagram where blocks are connected together.
- The components of this model pass information or data between one another. The output of one is the input to another. Inside those components can be simple equations, complex numerical models, or even Reduced Order Models (ROMs). These ROMs can be a complete complex 3D simulation that has been compacted down to the most essential equations that still accurately capture its behaviors but runs very quickly, sometimes even in real-time. This is a point of differentiation.
- This model is run as a simulation. Information and data can be continuously passed in this manner.
- An unusual capability, especially in the context of smart connected products and Digital Twins, is the ability to use sensor data in this model as an input. This is done by streaming the data to an IoT platform off of a connected physical product. This acts as just another input to one or several components.
- The model can be instrumented with virtual sensors, which are measurements taken from one or more of the components in the model. This augmented or enriches the sensor data that is being captured off the physical product. That provides more insight into the current or future operation of the product.
- Ultimately, want to be able to run this kind of model next to an IoT platform, or even at the edge on the physical product, so it can work against sensor data running in real-time. Twin Builder allows you to export this entire model as a software code. Here’s the critical point: (1) it is run headless and (2) runs independently of Twin Builder. That means it can be placed on an IoT platform or at the edge and will work.
I very rarely embed videos of demonstrations in reviews of products. However, the following one is stripped of market positioning and messaging while still communicates these capabilities.
There are several items to note in this specific demonstration.
- The Reduced Order Model is coming out of ANSYS’ Workbench. I expect many different kinds of ROMs from ANSYS offerings, and perhaps even others, will be pluggable into these models.
- Towards the end of the video, around the three-minute mark, the life expectancy predictions are produced with ANSYS Simulations within GE’s Predix platform.
- The context of this demonstration shows the value of Digital Twins. In this case, GE can validate that there is, in fact, an issue with the motor without sending a technician out on an expensive task for verification.
Commentary and Analysis on Twin Builder
What are the takeaways here?
A Fairly Complete Solution
Let’s be frank here. There are existing solutions that will allow you to build a 1D simulation. There are some that can use Reduced Order Models as a component. There are some that will accept streaming sensor data as input. There are some that export such models as code. However, there are very few that will do all of these things. Twin Builder can do all of those things today. That manifests in an easy to use solution where Digital Twins can be built fairly easily.
IoT Platform Integration
Whether you see this as an advantage or disadvantage depends on your position on integrations.
ANSYS does not offer an IoT platform. To deploy a digital twin from Twin Builder, you’ll likely need to find a way to ‘make it work’ either with some of your own ingenious employees or consultants for an initial setup. ANSYS has announced partnerships with PTC and GE. I assume that this includes some closer integration with their respective IoT platforms Thingworx and Predix.
Some look at this fact and rejoice, as it means they are independent and can work with pretty much any IoT platform. There is a reasonable argument that flexibility in determining which of your systems work with other systems is key. Alternatively, some look this reality and think that a lot of manual effort will be required to stitch everything together.
Your mileage may vary on this issue, but recognize the reality of the situation. Go into it with eyes wide open.
Over time, I expect the role of simulation to only rise in the enterprise because of Digital Twins. When executives realize the impact they can have on many different stages of the product lifecycle, they’re going to demand new initiatives to develop them.
Therein lies a major advantage of a company like ANSYS. You see, they have an incredibly broad simulation offering. Workbench provides broad and deep coverage of mechanical engineering physics. ANSYS has an array of simulation solutions for electromechanical, electronic and electrical systems. Their SCADE solution strongly supports embedded software development, connecting to Twin Builder for the plant side of 1D simulation. Altogether, they have a suite of analysis technologies that, when placed alongside Twin Builder, provide a pretty compelling proposition: developing incredibly accurate Digital Twins. That fact simply can’t be ignored.
Summary and Conclusions
- This week, ANSYS released Twin Builder, a solution to allow companies to build Digital Twins of their products.
- The solution provides capabilities to build 1D models of products and systems, integrate Reduced Order Models as components, use streaming sensor data as inputs, track measurements as virtual sensors, and export the model as software code that can be run next to an IoT platform or at the edge.
- The relative completeness of Twin Builder manifests in an easy-to-use solution. If ANSYS offered an IoT platform, deploying such models would be easier, but pioneering companies can get around that issue.
- The breadth and depth of ANSYS’ simulation solutions can’t be overlooked when assessing the value of Twin Builder. Customers will be able to leverage solutions for all those engineering physics domains to create Reduced Order Models, making the resulting Digital Twins more accurate.
Folks, those are my thoughts. What are yours? Interested to get your thoughts.