Integrate Generative Design into Your Alias Workflow

JAMES CRONIN: Hello, I'm James Cronin, Sales Development Executive at Autodesk. I'm joined by James Neville, Principal Business Consultant at Autodesk. Today we are going to discuss how to integrate generative design into your Alias workflow.

This new workflow is an alternative product development process. There are some common issues in the traditional designer engineer relationship. Often, the design process starts without engineering input. When engineering happens late in the process, this can lead to significant changes required of the design to maintain the performance targets.

This is an inefficient communication cycle that often leads to frustration on both sides. The ideal solution is one where design intent is maintained throughout the engineering refinement. The project adapts to the engineering requirements to achieve aesthetic goals, and one where the confirmation of engineering feasibility of the desired theme happens from the beginning of the project.

The process we're going to talk about today is called Design Intent Generative, or DIG. DIG is a design methodology and product development process with a focus on maintaining design intent while achieving upfront engineering feasibility. The benefit is a streamlined collaboration between design and engineering to create outcomes that simultaneously meet the requirements for aesthetic performance.

The example we'll be showing today was part of a research project that was run by Ford Performance in collaboration with Autodesk. The Ford Performance team has the responsibility to research new technologies and test new processes. They then define and productionize these workflows. The goal of this project was to investigate a blended workflow between Alias and Fusion 360 for product design. Design Intent Generative.

The Ford project team included Stavros Melabiotis and Dean Carbis from design, Dave Friske, Mike Niksa, and Marc Simon from engineering and research. In the example today, the Fusion generative design work was completed by Mike Niksa, and the final Alias design refinement was performed by Stavros.

This is the DIG workflow that was followed during the project that we will be covering today. The project starts in Alias, a quick model is created that is handed off to engineering early in the process. The data goes into Fusion 360, where the feasibility is developed by the engineer and the generative design output is created.

The project then goes back into Alias where that data is refined to achieve aesthetic goals. Once completed, the design is taken back into Fusion 360 to validate that the design edits still maintain the performance targets.

The first step in the process is to define the theme in Alias. Without guidance, Fusion 360 will build uninstructed outcomes, outcomes that perform well but do not follow a path that works with the design theme or brain character. So to control the results of generative design, the creative designer uses fast form finding techniques in Alias to build a simple model that will be the basis for the generative shape creation.

In this example, the design direction was for a split five spoke wheel. A simple model was created that defined the front facing silhouette of the wheel. This defines the theme direction for the design, the overall width of the spoke, the shape of the profile, the size of the split, as well as where the split starts and stops.

This build volume is driving the design intent. The model also incorporated basic engineering requirements of a wheel. The rim section, the hub mounting face, the bolt pattern, as well as the brake clearance requirements.

The process uses basic Alias build techniques. Positional continuity is all that is needed, as the goal is to create a volume or generative design will build within. The goal is to define the preserves and obstacles that will be used inside a Fusion 360, preserves or geometries that are required, like the rim and the center hub, or any other design element, obstacles or areas to avoid, like the geometry that defines the spoke split.

For more complex projects, the design intent could be developed using other Alias surfacing techniques, such as subdivisional surfaces or even Dynamo.

Now that the model is complete, we can prep it for Fusion handoff by performing the stitch function inside of Alias. Fusion 360 can open many CAD formats, but for an Alias designer, the easiest is to just save out the native Alias wire.

JAMES NEVILLE: Let's begin by talking about the overall strategy for generative design as it pertains to the development of this wheel. We have multiple goals to consider in this situation. First, we would like to use generative design to produce wheel spoke options that are aesthetically pleasing and fit within a certain design theme.

Second, we need a wheel that is performance focused and lightweight. Third and finally, the wheel must be structurally sufficient to withstand the rigors of spirited driving.

Now we'll examine the steps involved to achieve these goals. Defining the available design space, applying what we know about load cases via forces and constraints, choosing study materials and potential manufacturing methods, and finally reviewing outcomes and choosing the best one for our needs.

We begin by importing the design space previously created in Alias. The wire file uploads seamlessly into Fusion and retains the detailed surfacing necessary to start defining our generative setup. One quick note, depending on the creation method in Alias, the surfaces may need to be stitched together to create solid geometry. This process is fast and easy with Fusion surface modeling.

The next step is to create the geometry which will be used as preserve bodies in generative design. Preserves are regions of solid geometry that will be retained in the final outcome, and are typically located where structural loads are present, or where the part will be fixed to something else.

In the case of our wheel, the barrel must be retained because it forms a seal with the tire and is necessary to hold pressure. And it also transmits loads from the tire through to the rest of the wheel and suspension.

The hub must also be retained, because it is what we mount to the car. These geometric entities can be created from Scratch and Fusion with solid modeling, or can be extracted from the imported Alias geometry as shown here.

Obstacle geometry is necessary to confine the generative design process. Some generative design studies involve very unrestricted design spaces. A shelf bracket, for instance, cannot interfere with the wall or the shelf, but may occupy almost any other design space. Conversely, this generative wheel project has a very well defined design space.

The potential build space created in Alias dictates the region in which the spokes have freedom to develop. We need to assign obstacle geometry to the spaces that are off limits, and there are several ways to do this in Fusion. Since we have such well defined Alias geometry forming the foundation of our study, we can simply use the fluid volume feature to build a solid part around the barrel, hub, and spoke build volumes.

This feature automates the process of making a solid part, and then performs a Boolean subtract with the wheel parts as the tool bodies. The end result is a single, solid obstacle part that will influence generative design to create the forms only in the desired locations.

Now that we have our design space created, we need to tag the various entities in the generative design workspace. Preserves and obstacles are necessary here, while using a starting shape is optional. For ease of setup, preserves are turned green, while obstacles are turned red.

Applying real world loading scenarios is one of the critical steps in defining the generative design setup. These are typically gathered from the CAE team, and are used to stress a part in a simulation environment to evaluate overall strength, stiffness, durability, and predicted failure modes.

Relevant loads for this generative design setup include bending moments, radial forces, and impact loads. The procedure for assigning loads and constraints here is identical to that found in the simulation workspace.

By default, generative design will attempt to create outcomes with the lowest mass possible, without exceeding a specified safety factor. For our setup, we've chosen a safety factor of three, to maintain a margin of error with respect to maximum part stresses.

Generative design incorporates manufacturing criteria into the optimization process as well, attempting to create outcomes that are best suited for specific methods. We've chosen to examine both unrestricted outcomes and additive outcomes.

Finally, a custom material was created to match the aluminum chosen by Ford Performance. After the generative design process completes, a collection of outcomes are available to explore and compare. These outcomes may have variations in geometry depending on their manufacturing method, material choice, active load cases, and objectives and limits.

The Explore environment allows outcomes to be filtered and weighted based upon a range of preferences. The outcome we selected exhibited a good balance of strength and weight, while remaining aesthetically interesting to the design team. In the generative design Explore environment, the option to create a new design from this outcome was selected, establishing a fully editable parametric BREP part.

Now that a structurally sound generative design part has been created, there may be a desire to refine the surfacing that defines the organic parts of the wheel. While tools exist to handle engineering focused refinement in Fusion through T-Splines form body editing, a better route would be to bring the generative bodies back into Alias and into the hands of the design team. An OBJ control frame can be exported from the new design to accomplish just that.

JAMES CRONIN: The next step is to bring the generative design control frame into Alias for refinement. The goal is to take the generative output and refine and finesse the shapes to achieve the aesthetics that match the project. The designer also has the ability to add in details to the surfaces and turn the results into a compelling piece of art, while still achieving the weight and performance targets.

Generative design results import into Alias as subdivision data. This has benefits for the designer. The data is always continuous. This allows you to modify the data without having to worry about curvature or change in continuity.

The SubD limit services are represented by a set of NURBS surfaces. In contrast, Maya's limit surface is a polygon mesh. The NURBS surfaces have engineering benefits that will be covered during later refinement phases.

To bring the data into Alias, we are going to go File, Import, Subdiv. Alias can open an OBJ, but that won't give us the control we want over the subdiv data.

The next step is to isolate half of one spoke. Because the output from Fusion is the whole wheel, we need to isolate the one spoke by deleting the unused SubD faces, and then align the spoke across the two axes of symmetry.

All the SubD tools inside of Alias can be used to manipulate the data, but for this project we only used a handful of tools. The benefit of Alias's SubD limit surface being NURBS allows us to incorporate a hybrid workflow, which allows us to use traditional Alias modeling tools in conjunction with the SubD data to add in our engineered details.

The first tool to clean up the irregularities is this Align subdivision CVs to curves. This tool automatically generates a curve that can be used to then control the SubD cage. In this example, the tool was set to generate a degree two curve and create history. Also, to generate the curve.

This allows the data to be refined easier. Instead of having to worry about a whole series of CVs, I can just worry about a single, simple curve. And by manipulating that curve, it then updates the SubD geometry.

Keeping the CV layout clean and regular will help drive smooth shapes and forms. So another excellent tool is the SubDiv Planarize Edges tool. In this example, I'm using pick edge loop. Best fit is set to best fit plane, and then adaptive. This way I can quickly click on an edge loop and planarize it and clean up my data, so that I have more control in making sure that the highlights are going exactly where I want them to go.

The Crease SubDiv Edge tool allows you to quickly select some edge loops, and with a single click, turn those edge loops into a sharp edge. When combining this with the Align CVs to Curve tool, you can really control what the model will look like and add design elements, character lines quickly and easily into the data.

One of the main differentiators and powerful aspects of Alias's SubD modeling is the hybrid modeling. This allows you to incorporate engineering surfaces into your SubD model. In this example, the lug nut bolt hole location needs to be intersected with a SubD model. I can easily take my subdivision data and use Intersect and Trim, and still continue to manipulate my SubD data. And then the history will continue to update the trim based on the changes I've made to my SubD model.

Another hybrid modeling technique is using the Align tool. Because the SubD model is a series of NURBS surfaces, I can take those surfaces-- those individual patches-- and in this example align them to the center cap hole and maintain a perfect circular shape and still have complete control over my SubD cage.

Now that the refinement is complete, we can see the changes that were added to the model by the designer. So he maintained our design intent from our original silhouette theme profile that we created at the beginning, incorporated our generative design results, and then tuned them using some simple, easy to use SubD tools to clean the shapes, refine the volumes and the forms to focus on the aesthetic output.

The next step is to hand it back into Fusion to validate that the changes that were made still align with the performance targets that were initially set up.

JAMES NEVILLE: Now that Alias refinement of the generative design outcome has taken place, we will want to bring the final design back into Fusion to validate that it still meets functional requirements. Again, the Wire file is uploaded directly into Fusion, and surface modeling can be used to solidify any surface bodies.

In the simulation environment, the same load cases used to define the original generative design study are applied to the refined design, and a linear static stress analysis is performed. We use this validation environment to closely examine the wheel for stress concentrations, excessive displacement, and we want to examine the impact of the latest Alias refinements.

Comparing the wheel pre- and post- Alias refinement is quick and easy, and identifies the impact of aesthetic influence on part design, the impact of form on function. Exaggerated displacement animation shown here indicate how the wheel will flex while under a variety of loads.

Due to the generative design process, stress concentrations occur in a controlled and predictable manner while part weight is kept to a minimum, all while maintaining heavy influence from the design team. This final handshake between design and engineering truly shows the power of design intent generative process.

JAMES CRONIN: And now for the final step, let's bring the data into VRED to do some quick visualizations. It's a pretty simple model, only has two materials. So I'll quickly assign a metallic car paint and a carbon fiber, and then set up my renderings.

The first is a animation showing the wheel revolving. So we can sit and look at the forms and see how the light and reflections play across the shapes as it rotates. And here are some detail shots.

Thank you for watching the presentation.