Experimental Solvers: New Capabilities in Fusion 360 Generative Design

Share this Article

Autodesk is well known for having robust and reliable simulation tools. On the other hand, we have generative design in Fusion 360 software—the technology that revolutionizes the engineering world and the ways we design things. Autodesk development teams have been working on capitalizing on these proven simulation solvers as well as the latest technologies in generative design. As a result, a new tech preview feature named Experimental Solver is enabled in Fusion 360.

This means every user of the generative design feature will also see new outcomes with new sophisticated shapes, in addition to the existing traditional solutions. This technology brings new functionalities and delivers forms optimized in a new way, opening doors that have been closed previously. In addition to showing how to use the new technology of generative design in Fusion 360, above all, we will present how the new functionalities empower the new applications using real-world examples.

Related: Fusion 360 Introduction to Generative Design with Robert Savage

What Are Experimental Solvers in Generative Design?

Experimental Generative Solvers and Features is a preview feature available in Fusion 360. This is a technology which combines all abilities of generative design (advanced optimization algorithms, manufacturing constraints, design permutations) supported by new features (enforcing symmetry in outcomes, voids insertion in shapes) with mature, reliable simulation tools (advanced physics, buckling analysis, displacements limitation, frequency constraints, removing rigid modes). Additionally, Experimental Solvers and Features improves the robustness and performance of generative design solutions.

How to Use New Experimental Solvers in Fusion 360

The Experimental Solvers feature is a part of generative design which is available to every Fusion 360 user. To open the Generative Design Workspace inside Fusion 360, you need to navigate to the Generative Design option (2) inside the workspace selector (1) in the upper left corner of the Fusion 360 window.


As the Experimental Solvers feature is still being developed and improved, it is available as a preview feature. This means that to use it, you will need to enable it in the Preferences dialog box of Fusion 360. To do this, first open the Preferences window by choosing the  Preferences option (4) inside the menu (3) in the upper right corner of the Fusion 360 window.


Inside the Preferences window, find the Preview Features option (5) and then make sure that under the Generative Design tab you have selected Experimental Generative Solvers and Features (6).


The above steps will enable the Experimental Solvers Tech Preview for your account. Now to explore alternative outcomes for a specific study, one more step is needed. After creating a new study, open its Study Settings (7,8) and select the Alternative Outcomes option (9).



When you run the study with the Alternative Outcomes option selected, you will be able to use new features that are described in more detail later, and you may get some additional outcomes on top of the default ones.

If you are a new user of generative design and have no knowledge of how to create or run a generative design study, reference Fusion 360 Introduction to Generative Design.

Alternative Outcomes

Thanks to the continuous work on design divergence, new Experimental Solvers offer up to four additional outcomes, plus the default ones, for every study setup. The final number of outcomes may vary depending on the chosen study configuration; that is, manufacturing constraints, additional constraints like symmetry, etc.

Please note that Experimental Solvers is still a preview feature, which means that some of the new features may not be working with all manufacturing constraints and final outcomes count might be smaller. Some of the configurations are not possible to run at all. The full list of supported manufacturing methods is presented later, next to the description of every new constraint.


After opening the results in the Explore tab, you can identify the outcomes that were generated by Experimental Solvers by the flask icon in the upper right corner of the outcome thumbnail.

Capabilities: New Design Variants

One of the characteristics of the new Experimental Solvers is increased variety of shapes. Thanks to new ways of optimization, the results you get for the same study setup can differ significantly in terms of the form while continuing to meet the optimization objectives. This gives you the freedom to choose the design based on its aesthetic values and not only mechanical properties.

Below are presented some examples that show this behavior. Both design variants have remarkably similar parameters, like a minimum factor of safety or mass, but are completely different visually.





Capabilities: Shapes with Voids

Another interesting feature of Experimental Solvers is its ability to create void shapes. In the standard approach of the generative design solver, the material can only be removed from the outer surface of the model. In most cases, this method provides satisfactory results; however, there are some situations where it is insufficient, and the outcomes that it produces are not the optimal ones.


The improved algorithms of Experimental Solvers provide the possibility to remove the material also from the inside if it makes sense for a given model. This allows for faster optimization in some of the cases and produces the shapes that are more optimal from an engineering point of view.


Let’s look at the example. The picture below shows the beam being subjected to a torque load. The green parts are preserve geometries while the yellow part is a starting shape. 


When you run the static stress analysis on this model (in the Simulation workspace inside Fusion 360, for example) you will see that maximum stresses are located on the outer surface of the model, while the inside of the model is subjected to much lower stress values.

The most logical and most optimal solution here would be the shape that has its outer edges preserved and is empty inside. However, up until now, the generative design solver was unable to produce such a shape. The Experimental Solver brings new possibilities, and you can see the results in the image below. At the top of the image is the design produced by the standard generative design solver. At the bottom, there is an outcome created by a new optimization method used in Experimental Solver. The second design is much closer to the one that is indicated by stresses distribution inside of the model. It is also much lighter and has a much lower displacement value than the traditional design.


Capabilities: Displacements Limitation


Displacements are directional motions of the structure under applied loads. Apart from the load-bearing capacity requirements, many design standards also limit the maximum allowable displacements.

In the traditional generative design approach, we can see the maximum displacement of the generated outcome, but we are unable to directly impact it before the post-processing stage.

The example of a considered generative design outcome.
Parameters of the generated outcome.
Parameters of the generated outcome.
Simulation results on the outcome for displacements.

With Experimental Solvers and Features, we can gain more control by constraining allowable displacements. As the preview option is turned on, the additional options appear in Objectives and Limits.

Global Displacements

The first constraint is related to global displacements, which are total motions of nodes in a model considered in a global coordinate system.

Model Setup

1. On the Define tab, click Design Criteria > Objectives
2. Click Displacement and select Global
3. Choose from one up to three directions
4. Specify the maximum displacement value for X, Y, and/or Z directions

Setup for constraining global displacements.

To limit global displacements in the considered model, we set the allowable maximum value 0.1 mm in all three orthogonal directions.

Then we compare two final outcomes: without displacement constraints and with global constraints. As the result, we can see that the final outcomes have shape optimized in a similar way. For the one with limited displacements, there are fewer but more stocky members.

Comparing these properties, we can see that the Factor of Safety for both is the same. However, maximum global displacement was reduced from 0.28 mm to 0.1 mm as it was constrained.

The outcome without displacement constraints.

Want more? Download the full class handout to read on.

Krystian Motawa graduated with a master's degree in Structural Engineering. He also completed postgraduate studies in software development methodologies. Prior to Autodesk, he worked as a structural engineer. He joined Autodesk in 2016, and is currently working as a senior software engineer. He's been involved in generative design from the beginning of this project and is based in Krakow, Poland.

Grzegorz Borowski is an Autodesk employee with four years of service. He is currently a product owner working mainly on generative design. He graduated with a master's degree in Mechanical Engineering, specializing in Finite Element Method (FEM) simulations. Prior to joining Autodesk, he worked as a CAE engineer for several companies in the automotive industry.

Karolina Czechowicz is a software engineer with a research background in materials engineering and simulation. She graduated in Applied Computer Science with a specialization in Modeling and Information Technologies. An Autodesk employee for three years, she is currently working on the generative design solver.