Unlocking Better Engineering with Generative Design

CDFAM Speaker Series Interview with Mike Smell, Senior Product Manager, Fusion 360, Autodesk

Autodesk has been at the forefront of raising awareness, and democratizing access to Generative Design software for engineering via the Fusion 360 platform.

Initially emerging from the Project Dreamcatcher as part of Autodesk’s research division, generative design is now available to all users of Fusion 360 as a cloud based extension.

Senior Product Manager for Fusion 360 at Autodesk, Mike Smell, will be showcasing the latest developments in Autodesk’s Generative Design at CDFAM, as well as shedding light on how the adoption of generative design can accelerate and improve advanced engineering applications.

Ahead of his presentation at CDFAM, we asked Mike a few questions about how generative design aligns with Autodesk’s broader manufacturing strategy, some of the advantages and misconceptions he encounters from customers engaging with this technology and how they can best get started adopting generative design into their engineering workflow.

With Autodesk’s history of democratizing design software across various fields, from large scale urban planning and architecture, through to mechanical design, and electronics, from the early days of AutoCAD revolutionizing drafting, through to TinkerCAD enabling STEM education, how does the addition of Generative Design in Fusion 360 support and enhance this fundamental product philosophy or strategy?

Autodesk recently announced the establishment of three industry-specific cloud platforms to address the needs of our customers, for our Design & Manufacturing customers this industry cloud is called Autodesk Fusion.

Within the Autodesk Fusion industry cloud, we have Fusion 360, the delivery vehicle that connects our customers, their data, and their network across the entire product development life cycle. Fusion 360 provides professional tools for 3D modeling, CAD, CAM, CAE, and PCB product design and manufacturing. 

Generative design technologies in Fusion 360 are built for the engineer, with ease of use in mind, and integrated into the product development process. We incorporated these technologies within Fusion 360 to ensure every engineer can access these capabilities as part of their design process instead of reserving them for a separate standalone application or specialist audience. 

We strive to ensure that Fusion 360 users can easily extract the value of design exploration, improved productivity, and better engineering outcomes from the start of the design process.

The term ‘generative design’ is sometimes debated and compared to other concepts like computational, algorithmic, and simulation-driven design or topology optimization. In the context of Autodesk software, how is ‘Generative Design’ defined, and what distinguishes it from these other terms and approaches?

At a high level, Autodesk views generative design as intelligent design exploration technology that generates multiple actionable outcomes based on user-supplied requirements.

These design outcomes enable engineers to compare and make tradeoff decisions based on what best meets their needs.

We strongly believe that generative design should be viewed as complementary to the design process, enabling improved productivity and better engineering outcomes.

If we take a closer look at the generative design technologies in Fusion 360, we offer solutions for different stages in the process.

Automated Modeling in Fusion 360

Early on, Automated Modeling lives right in the Design workspace and provides multiple design alternatives with minimal inputs in minutes. When users are ready to dive further into engineering, validation, and optimization of designs, the Generative Design workspace provides solutions for structural and fluid flow applications, producing designs for a specified set of requirements.

Our generative design technologies are different in how we approach geometry inputs, combine physics, manufacturing, and production cost all into one approach, and produce editable CAD results.

As an example, simulation-driven and topology optimization often require an existing design or design space to get started, where generative design can work with only critical geometry for design intent and generate the rest of the geometry to connect everything, making it easier to challenge that status quo for a given design.

On the other side, computational and algorithmic approaches don’t always include concepts of physics or manufacturing, meaning there could still be guesswork in the effectiveness of the design.

What are the indicators or criteria that can help designers and engineers determine when it is the optimal time to utilize generative design for addressing their engineering challenges?

Generative design makes a lot of sense when engineers are pursuing new product proposals, design transformation, or are working to develop solutions that support the manufacturing or assembly process. Great manufacturing examples are robotic end effectors or jigs and fixtures for complex designs. 

Business goals around improving product performance, increasing engineering productivity and innovation capacity, winning more business, or reducing product cost or weight, are also great opportunities for generative design. 

From a more technical point of view, if a design goes through simulation workflows for stress, vibration, and/or buckling, or you have the flexibility to use a different material or manufacturing process, there is also a great chance that generative design can help you find a better engineering outcome.

Anytime performance, manufacturing and business objectives need to be balanced in the design process, an intelligent system like generative design can expand the exploration of design alternatives to improve the decision-making process.

What are some of the most common misconceptions or unrealistic expectations you have encountered from individuals seeking to employ generative design for addressing engineering challenges, and how do you help clients overcome them?

One of the most common issues we encounter is user expectations.

We often find that users aspire to have generative design replace traditional design and simulation workflows. They expect to be able to define requirements for all the potential physics and validation studies required to bring a design to market, creating an all or nothing situation. 

In these scenarios, we work with our customers to integrate generative design to complement their existing process and help them unlock the value of generative design as a collaborator and accelerator in the design process, even without designing for all their potential requirements or replacing their entire process.  

Another common challenge is when users are limited in how far they can explore for a given design. For example, if they are limited to one material and manufacturing method, a more traditional optimization or design of experiments workflow may be a better option.

The outcomes generated by algorithmic or generative design are highly dependent on the quality of inputs and the clarity of problem definition. How does Fusion 360 assist engineers in framing the right questions, and does it provide any warnings or guidance when they might not be on the right track?

Generative Design in Fusion 360 gives the user guidance on the principles of the design space setup, however, it doesn’t give guidance on the definition of performance requirements or the accuracy of how they represent a desired application. The user must understand the application and translate it into load cases for the study. 

Users who are less familiar with the physics of their application or are not ready to spend all the time defining those requirements, might be better starting with Automated Modeling to explore shapes and solutions to their design challenges.  

How has the integration of machine learning and artificial intelligence influenced the development and capabilities of generative design, and what potential advancements can we expect in the near future?

Generative Design in Fusion 360 currently leverages a machine-learning-based approach for identifying and sorting outcomes by visual similarity.

Visual similarity using machine learning in Fusion 360

With the rapid advance of AI technologies in recent months, the possibilities are nearly endless, and we are aggressively researching how they can best move the needle in the product development process.

You can be sure that advancements in this area will continue to focus on helping our users be more productive and achieve better engineering outcomes. A great potential opportunity would be helping translate the performance requirements for the design into correct load cases for the study.

Fusion 360 combines parametric CAD, generative design, surface modeling, as well as subtractive and additive manufacturing preparation, simulation, and execution, empowering designers, engineers, and teams to design and manufacture within the same software platform. In cases where other software or hardware tools are necessary in the workflow, how is data transferred in and out of Fusion 360 to ensure seamless integration?

Fusion 360 provides AnyCAD workflows to establish an associative link for incoming 3rd party data that can be used for a Generative Design study.

A resultant generative design outcome is a fully editable Fusion 360 model. Fusion 360 can export to many different 3rd party CAD or neutral formats for use in other systems. Autodesk Inventor users have a specific workflow to help isolate design geometry and use it associatively with Fusion 360 for generative design.

To wrap up, what are the main takeaways you aim to impart through your presentation at the CDFAM symposium, and what insights or learnings do you hope to acquire during the event?

First, Fusion 360 provides professional tools for 3D modeling, CAD, CAM, CAE, and PCB product design and manufacturing that connects users, their data and their networks across the entire product development lifecycle. We have made our generative design technologies highly accessible to help users achieve better engineering outcomes. 

Generative Design is an intelligent collaborator and accelerator in the product development process. Generative design should not be seen as a replacement for existing processes or people but is well suited to help users improve engineering productivity, improve product performance, and reduce costs. 

Lastly, generative design technologies in Fusion 360 have helped users deliver real results, through improved productivity and better engineering outcomes as they strive to solve today’s most challenging problems. 

As for me, I hope to learn how others perceive the value of generative design and other automated design technologies and gain a broader view of how they are impacting the world of product design. I am looking forward to networking with all the brilliant individuals who are working in a similar technology area.

Recent Posts