Four years ago, I wrote a series of posts with the common theme: The road to model-based and connected PLM. I discussed the various aspects of model-based and the transition from considering PLM as a system towards considering PLM as a strategy to implement a connected infrastructure.

Since then, a lot has happened. The terminology of Digital Twin and Digital Thread has become better understood. The difference between Coordinated and Connected ways of working has become more apparent. Spoiler: You need both ways. And at this moment, Artificial Intelligence (AI) has become a new hype.

Many current discussions in the PLM domain are about structures and data connectivity, Bills of Materials (BOM), or Bills of Information(BOI)  combined with the new term Digital Thread as a Service (DTaaS) introduced by Oleg Shilovitsky and Rob Ferrone. Here, we envision a digitally connected enterprise, based connected services.

 

A lot can be explored in this direction; also relevant Lionel Grealou’s article in Engineering.com: RIP SaaS, long live AI-as-a-service and follow-up discussions related tot his topic. I chimed in with Data, Processes and AI.

 

However, we also need to focus on the term model-based or model-driven. When we talk about models currently, Large Language Models (LMM) are the hype, and when you are working in the design space, 3D CAD models might be your first association.

There is still confusion in the PLM domain: what do we mean by model-based, and where are we progressing with working model-based?

A topic I want to explore in this post.

 

It is not only Model-Based Definition (MBD)

Before I started The Road to Model-Based series, there was already the misunderstanding that model-based means 3D CAD model-based. See my post from that time: Model-Based – the confusion.

Model-Based Definition (MBD) is an excellent first step in understanding information continuity, in this case primarily between engineering and manufacturing, where the annotated model is used as the source for manufacturing.

In this way, there is no need for separate 2D drawings with manufacturing details, reducing the extra need to keep the engineering and manufacturing information in sync and, in addition, reducing the chance of misinterpretations.

MBD is a common practice in aerospace and particularly in the automotive industry. Other industries are struggling to introduce MBD, either because the OEM is not ready or willing to share information in a different format than 3D + 2D drawings, or their supplier consider MBD too complex for them compared to their current document-driven approach.

In its current practice, we must remember that MBD is part of a coordinated approach.

Companies exchange technical data packages based on potential MBD standards (ASME Y14.47 /ISO 16792 but also JT and 3D PDF). It is not yet part of the connected enterprise, but it connects engineering and manufacturing using the 3D Model as the core information carrier.

As I wrote, learning to work with MBD is a stepping stone in understanding a modern model-based and data-driven enterprise. See my 2022 post: Why Model-based Definition is important for us all.

To conclude on MBD, Model-based definition is a crucial practice to improve collaboration between engineering, manufacturing, and suppliers, and it might be parallel to collaborative BOM structures.

And it is transformational as the following benefits are reported through ChatGPT:

  • Up to 30% faster in product development cycles due to reduced need for 2D drawings and fewer design iterations. Boeing reported a 50% reduction in engineering change requests by using MBD.

  • Companies using MBD see a 20–50% reduction in manufacturing errors caused by misinterpretations of 2D drawings. Caterpillar reported a 30% improvement in first-pass yield due to better communication between design and manufacturing teams.
  • MBD can reduce product launch time by 20–50% by eliminating bottlenecks related to traditional drawings and manual data entry.
  • 20–30% reduction in documentation costs by eliminating or reducing 2D drawings. Up to 60% savings on rework and scrap costs by reducing errors and inconsistencies.

Over five years, Lockheed Martin achieved a $300 million cost savings by implementing MBD across parts of its supply chain.

 

MBSE is not a silo.

The MBSE playground

For many people, Model-Based Systems Engineering(MBSE) seems to be something not relevant to their business, or it is a discipline for a small group of specialists that are conducting system engineering practices, not in the traditional document-driven V-shape approach but in an iterative process following the V-shape, meanwhile using models to predict and verify assumptions.

And what is the value connected in a PLM environment?

A quick heads up – what is a model

A  model is a simplified representation of a system, process, or concept used to understand, predict, or optimize real-world phenomena. Models can be mathematical, computational, or conceptual. 

We need models to: 

  • Simplify Complexity – Break down intricate systems into manageable components and focus on the main components.
  • Make Predictions – Forecast outcomes in science, engineering, and economics by simulating behavior – Large Language Models, Machine Learning. 
  • Optimize Decisions – Improve efficiency in various fields like AI, finance, and logistics by running simulations and find the best virtual solution to apply.
  • Test Hypotheses – Evaluate scenarios without real-world risks or costs for example a virtual crash test..

It is important to realize models are as accurate as the data elements they are running on – every modeling practices has a certain need for base data, be it measurements, formulas, statistics.

I watched and listened to the interesting podcast below, where Jonathan Scott and Pat Coulehan discuss this topic: Bridging MBSE and PLM: Overcoming Challenges in Digital Engineering. If you have time – watch it to grasp the challenges.

The challenge in an MBSE environment is that it is not a single tool with a single version of the truth; it is merely a federated environment of shared datasets that are interpreted by modeling applications to understand and define the behavior of a product.

In addition, an interesting article from Nicolas Figay might help you understand the value for a broader audience. Read his article:  MBSE: Beyond Diagrams – Unlocking Model Intelligence for Computer-Aided Engineering.

Ultimately, and this is the agreement I found on many PLM conferences, we agree that MBSE practices are the foundation for downstream processes and operations.

We need a data-driven modeling environment to implement Digital Twins, which can span multiple systems and diagrams.

In this context, I like the Boeing diamond presented by Don Farr at the 2018 PLM Roadmap EMEA conference. It is a model view of a system, where between the virtual and the physical flow, we will have data flowing through a digital thread.

Where this image describes a model-based, data-driven infrastructure to deliver a solution, we can, in addition, apply the DevOp approach to the bigger picture for solutions in operation, as depicted by the PTC image below.

 

Model-based the foundation of the digital twins

To conclude on MBSE, I hope that it is clear why I am promoting considering MBSE not only as the environment to conceptualize a solution but also as the foundation for a digital enterprise where information is connected through digital threads and AI models (**new**)

The data borders between traditional system domains will disappear – the single source of change and the nearest source of truth – paradigm, and this post, The Big  Blocks of Future Lifecycle Management, from Prof. Dr. Jörg Fischer, are all about data domains.

However, having accessible data using all kinds of modern data sources and tools are necessary to build digital twins – either to simulate and predict a physical solution or to analyze a physical solution and, based on the analysis, either adjust the solutions or improve your virtual simulations.

Digital Twins at any stage of the product life cycle are crucial to developing and maintaining sustainable solutions, as I discussed in previous lectures. See the image below:

Conclusion

Data quality and architecture are the future of a modern digital enterprise – the building blocks. And there is a lot of discussion related to Artificial Intelligence. This will only work when we master the methodology and practices related to a data-driven and sustainable approach using models. MBD is not new, MBSE perhaps still new, building blocks for a model-based approach. Where are you in your lifecycle?