Generative Design: Insights, Myths, and More
Generative design is an iterative process that offers engineers a way to create and optimize products in a way that is impossible with traditional methods. It is also frequently misunderstood, incorrectly lumped in with generative AI, and assumed to replace the role of the engineer. In this piece, we clearly define what generative design is, what it can do for you, and highlight how Tech Soft 3D supports generative design workflows.
What is Generative Design?
Generative design involves outlining a series of clearly defined criteria, or constraints, and using software tools to create outputs based on these parameters. This type of design process is iterative, with both the output and criteria being refined until the end product is satisfactory.
Generative design is used across a wide array of industries, from engineering, aerospace, and architecture to the design of consumer products and urban planning. Crucially, the term generative design does not refer to one set of technologies or single process, but a broad design process defined by its iterative nature, constraints, and deviation from more traditional design methods.
Generative Design vs Artificial Intelligence
There are many different types of AI. The current explosion in conversation on the topic is fueled in part because of developments in generative AI. As the name suggests, this type of artificial intelligence can, based on prompts, create images, texts, videos, code, and more based on their large language models. This "creative" aspect, along with sharing the word "generative" has led to plenty of confusion about generative AI versus generative design.
To be crystal clear: generative design is not a type of AI-based technology. While there are places in the generative design workflow where AI tools can be used, the process itself is not inherently AI-based, nor is AI widely used in engineering applications of generative design.
A much more useful comparison for developers, decision-makers, and end users is to discuss how generative design is different from traditional engineering.
Generative Design vs. Traditional Design
Generative design sees engineers spend far more time and energy than traditional methods outlining constraints before the creation of a design. The design is created using algorithms that are informed by these criteria and are then tested through simulation tools and optimized. The different stages of this process are repeated until the engineers are satisfied with the results. The three broad stages of the generative design process are:
Defining the constraints
Exploring the shape
Simulation and optimization
Constraint Definition
In the defining the constraints stage, engineers start by creating or adjusting the functional requirements and constraints; a huge amount of effort goes into this process. This stage also relies on establishing a clear definition of the goals of the design, the data that will be used to inform the algorithms, and the best algorithms to use.
Constraints can include physical and functional needs like thermal dynamics, structural requirements, and then materials, costs, availability of resources, etc. Additionally, manufacturing constraints are considered, such as the limitations of a CNC machine or how the cooling process in injection molding can create warping.
These requirements inform the generative design algorithm before the creation of a 3D model, instead of traditional simulation after a model is created.
The process of clearly defined rules that inform the design process is a cornerstone of generative design
Exploring the Shape
The next phase of the process is where a design of the object is created, using algorithms informed by the constraints established in the last step. The two common technologies used for this stage of generative design algorithms are based on topology optimization and parametric modeling. Many other techniques to explore the shape exist, but the CAD industry can ultimately only utilize ones that produce manufacturable shapes.
Most generative design technology on the market today utilizes topology optimization. Not dissimilar to how traditional manual CAD design often begins with a shape and sees material taken away, the process starts with a cube (or other shape) and material is removed from there.
A parametric modeler would take a different approach. This process starts with an initial parametric design. The design might include a hole with a radius of 5 cm. Exploring the shape would involve the algorithm exploring different sizes for that hole.
Simulation and Optimization
Engineers use a variety of simulation tools to evaluate the product of the algorithms in the previous steps. This will vary significantly depending on what you are creating.
The engineer uses optimization tools and the parameters they’ve outlined to iterate on different shapes. Engineers try to get the part as close to their ideal as possible and then repeat the previous steps to refine as needed. The underlying algorithm and constraints can be changed, with the engineer modifying the process to get close to the desired result.
Simulation plays a role throughout the generative design process. With the key constraints outlined and tweaked repeatedly, the algorithm simulates many designs and presents some that may fit the acceptance criteria. The algorithm itself is also adjusted. Just like in a traditional workflow, the results are used to iterate and improve the final result.
Often, the final product still needs work in traditional CAD software, a mesh smoothing tool, or another process to make a generated design more manufacturable.
Why Use Generative Design?
The generative design process is significantly different from traditional design processes and offers a variety of benefits to engineers. Traditional design requires different designs to be created and tested. Human input is required throughout the process. Arriving at the optimal solution relies on refining the model, guided by simulation, testing, and experience. Each iteration takes time and money, and teams can often manage only a handful of iterations in the time/budget available.
Generative design allows teams to potentially iterate thousands of times, testing each under a larger variety of conditions and constraints. As you can see in the example above, the results can be unlike something you would see a human designer create. This allows engineers to test far more ideas than humanly possible, thus arriving at more optimal designs that may be missed in more traditional workflows. Generative design, as a result of this efficiency, can save organizations significant time and money when used well.
Does Generative Design "Replace Engineers"?
No. Engineers remain essential to every stage of the design process, whether that is traditional or generative.
The creation of requirements, constraints, and contextual rules for the simulation model is one of the most complex parts of the new design process. Engineers must carefully boil down the problem into its core elements and constraints, an extremely difficult task.
Engineers are critical to modifying, or coaching, the simulation model and judging the output. After the model creates a design and evaluates it through its process, engineers must examine the results and determine what needs changing.
Despite its utility, the generative design process is extremely limited in its ability to create more complicated parts. While our simulation and constraint-building technology is powerful, these tools are not perfect representations.
For larger and more complex parts and assemblies, multiphysics simulations and testing yield far from fully accurate results. As a result, engineers must use generative design carefully, on individual parts or small assemblies, before using their own experience and other tools. The more detailed a component or the loads it will be subjected to, the more challenging the role of the engineer.
How Does Tech Soft 3D Support Generative Design?
Generative design is changing the way engineers work across a wide range of industries. This introduces new challenges and opportunities for developers. Our SDKs provide unique ways to address pain points, drive innovation, and get the most out of generative design and other creative processes. No matter what engineering software you are creating, Tech Soft 3D toolkits can offer you a competitive advantage while saving you time and money in the development process.
We can support your CAD data import needs with HOOPS Exchange for direct access to over 30 different CAD file formats, while CEETRON Access delivers API access to just as many CAE file formats, including simulation results and preprocessed meshes. These toolkits allow you to access rich 3D within your application and provide easy creation of learning sets to train an AI model or, as a prompt, a starting point for a generative algorithm.
Generative algorithms need tools to both define and solve simulation models. CEETRON Solve is a set of developer tools that jumpstart algorithm development and implementation with a diverse set of linear algebra and matrix tools and interactive linear and eigen solvers. Another way Tech Soft 3D is supporting the evolution of computer-aided design and manufacturing is through our powerful visualization toolkits.
Generative design algorithms can output many varying designs that need to be reviewed by an actual person. A reliable and powerful graphics framework for visualizing and navigating this information is crucial. Our toolkits empower developers to create intelligent, interactive 3D interfaces within their applications. Over half of today’s CAD and CAM systems are powered by Siemens Parasolid, the world’s leading solid modeling kernel. Tech Soft 3D is the sole reseller of Parasolid, and we have direct integrations with many of our existing toolkits, which allows for rapid application development. Solid modeling is important for BREP and parametric model creation and refinement.
HOOPS Visualize supports those building for desktop and mobile applications, while HOOPS Communicator supports developers building 3D web applications. For those working with CAE data, CEETRON Envision offers the complete package of CAE data import, analysis, visualization, reporting, and automation.
Alternatively, if your data model is based on polygons or you are building additive manufacturing (3D printing) software, we are the exclusive reseller of Machineworks Polygonica, the marketing leading polygonal and mesh modeling toolkit. As with Parasolid, we have direct integrations with many of our existing SDKs.
As mentioned, generative design is still in its formative stages. With improvements in technology, from simulation and analysis to more powerful hardware and artificial intelligence, generative design is only going to grow in its capacity to revolutionize design processes.
Tech Soft 3D is excited to support these changes and will continue to develop its capabilities in these ground-breaking technologies.
Get in touch to learn more about how our products can support your application development, no matter what your workflow is.
Got a question? Let one of our experts help you find the answer