Combustion engines, propellers, and hydraulic pumps are examples of fluidic devices, which are appliances that use fluids to perform certain functions, such as generating power or transporting water.
Because fluid devices are so complex, they are often developed by experienced engineers who manually design, prototype, and test each apparatus in an expensive, time-consuming, and labor-intensive iterative process. But in a new system, the user only needs to specify the positions and velocities where the fluid enters and exits the device – the computational pipeline then automatically creates the optimal design that achieves these goals.
The system could make it faster and cheaper to design fluid devices for any application, such as microfluidic labs on a chip that can diagnose disease from a few drops of blood, or artificial hearts that can save the lives of transplant patients.
Recently, computational tools have been developed to simplify the manual design process, but these techniques had limitations. Some required a designer to predetermine the shape of the device, while others represented shapes using 3D cubes, known as voxels, resulting in box-like, ineffective designs.
The computational technique developed by researchers from MIT and elsewhere overcomes these pitfalls. Design optimization frameworks do not require the user to make assumptions about how a device should look. And the shape of the device automatically evolves during optimization, with smooth, imprecise boundaries instead of blocky. This allows their system to create more complex shapes than other methods.
“You can now seamlessly perform all these steps in a computational pipeline. Also, with our system you can potentially create better devices because you can discover new designs that have never been explored before with manual methods. Maybe there are some shapes that have not been tried before. They haven’t been discovered by experts yet,” he says. Yifei Li, an electrical engineering and computer science graduate student, is the lead author of a paper detailing the system.
Co-authors include Tao Du, a former postdoc in the Computer Science and Artificial Intelligence Laboratory (CSAIL) and currently an assistant professor at Tsinghua University; and senior author Wojciech Matusik, professor of electrical engineering and computer science, who leads the Computational Design and Manufacturing Group at CSAIL; The University of Wisconsin at Madison, LightSpeed Studios, and Dartmouth College, among others. The research will be presented at ACM SIGGRAPH Asia 2022.
Shaping a fluid device
The researchers’ optimization pipeline begins with an empty, three-dimensional region divided into a grid of small cubes. Each of these 3D cubes or voxels can be used to form part of the shape of a fluid device.
One thing that sets their system apart from other optimization methods is how it represents or “parameterizes” these tiny voxels. Voxels are parameterized as anisotropic materials, which are materials that respond differently depending on the direction of the force applied to them. For example, wood is much weaker to forces applied perpendicular to the grain.
They use this anisotropic material model to parameterize voxels with properties such as all solid (as they would be found outside the device), all liquid (liquid inside the device), and voxels located at the solid-liquid interface. both solid and liquid material.
“When you’re going in the solid direction, you want to model the material properties of solids. But when you’re going in the fluid direction, you want to model the behavior of liquids. This inspired us to use it to represent anisotropic materials at the solid-liquid interface and allows us to model the behavior of this region very accurately,” he explains. Li.
Computational pipelines also think differently about voxels. Rather than simply using voxels as 3D building blocks, the system can angle the surface of each voxel and change its shape in very precise ways. Voxels can then be formed into smooth curves, enabling complex designs.
After their system creates a shape using voxels, it simulates how fluid flows through that design and compares it to user-defined targets. It then adjusts the design to better meet the objectives and repeats this pattern until it finds the most suitable shape.
With this design in hand, the user can use 3D printing technology to manufacture the device.
After the researchers created this design pipeline, they tested it with cutting-edge methods known as parametric optimization frameworks. These frames require designers to specify in advance what they think the shape of the device should be.
“When you make this assumption, all you get are variations of the shape within a family of shapes,” Li says. “But our framework doesn’t need you to make such assumptions because we have a very high degree of design freedom by representing this space with a large number of small voxels, each of which can change its shape.”
In each test, its frameworks outperformed baselines, creating smooth shapes with complex structures that could be too complex for an expert to predict. For example, it automatically created a tree-shaped fluid diffuser that bypassed an obstacle in the middle of the device while transferring fluid from one large inlet to 16 small outlets.
The pipeline also produced a vane-shaped device to create a swirling flow of fluid. To achieve this complex shape, their system automatically optimized about 4 million variables.
“I was really pleased to see that our pipeline was able to automatically develop a propeller-shaped device for this fluid bender. This shape drives a high-performance device. If you’re modeling this target with a parametric shape frame, because it can’t be. the device will not perform the same.”
While Li is impressed with the variety of shapes it can automatically generate, he plans to improve the system using a more complex fluid simulation model. This will enable the pipeline to be used in more complex flow environments, allowing it to be used in more complex applications.
This research was supported in part by the National Science Foundation and the Defense Advanced Research Projects Agency.